seqsero_v2: libs/mapping_and_assembly

author	cstrittmatter
date	Thu, 21 Dec 2017 12:45:31 -0500
parents
children

rev	line source
0 6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	1 import os,sys,glob,time,itertools,subprocess
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	2 from Initial_Conditions import phase1
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	3 from Initial_Conditions import phase2
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	4 from Initial_Conditions import phaseO
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	5 from Initial_Conditions import sero
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	6 from distutils.version import LooseVersion
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	7
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	8
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	9
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	10
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	11 def xml_parse_score_comparision_seqsero(xmlfile):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	12 #used to do seqsero xml analysis
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	13 from Bio.Blast import NCBIXML
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	14 handle=open(xmlfile)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	15 handle=NCBIXML.parse(handle)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	16 handle=list(handle)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	17 List=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	18 List_score=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	19 List_ids=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	20 for i in range(len(handle)):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	21 if len(handle[i].alignments)>0:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	22 for j in range(len(handle[i].alignments)):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	23 score=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	24 ids=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	25 List.append(handle[i].query.strip()+"___"+handle[i].alignments[j].hit_def)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	26 for z in range(len(handle[i].alignments[j].hsps)):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	27 if "last" in handle[i].query or "first" in handle[i].query:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	28 score+=handle[i].alignments[j].hsps[z].bits
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	29 ids+=float(handle[i].alignments[j].hsps[z].identities)/handle[i].query_length
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	30 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	31 if handle[i].alignments[j].hsps[z].align_length>=30:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	32 #for the long alleles, filter noise parts
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	33 score+=handle[i].alignments[j].hsps[z].bits
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	34 ids+=float(handle[i].alignments[j].hsps[z].identities)/handle[i].query_length
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	35 List_score.append(score)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	36 List_ids.append(ids)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	37 temp=zip(List,List_score,List_ids)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	38 Final_list=sorted(temp, key=lambda d:d[1], reverse = True)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	39 return Final_list
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	40
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	41
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	42 def Uniq(L,sort_on_fre="none"): #return the uniq list and the count number
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	43 Old=L
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	44 L.sort()
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	45 L = [L[i] for i in range(len(L)) if L[i] not in L[:i]]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	46 count=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	47 for j in range(len(L)):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	48 y=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	49 for x in Old:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	50 if L[j]==x:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	51 y+=1
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	52 count.append(y)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	53 if sort_on_fre!="none":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	54 d=zip(*sorted(zip(count, L)))
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	55 L=d[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	56 count=d[0]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	57 return (L,count)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	58
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	59
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	60 def judge_fliC_or_fljB_from_head_tail_for_one_contig(nodes_vs_score_list):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	61 #used to predict it's fliC or fljB for one contig, based on tail and head score, but output the score difference,if it is very small, then not reliable, use blast score for whole contig to test
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	62 #this is mainly used for
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	63 a=nodes_vs_score_list
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	64 fliC_score=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	65 fljB_score=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	66 for z in a:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	67 if "fliC" in z[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	68 fliC_score+=z[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	69 elif "fljB" in z[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	70 fljB_score+=z[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	71 if fliC_score>=fljB_score:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	72 role="fliC"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	73 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	74 role="fljB"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	75 return (role,abs(fliC_score-fljB_score))
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	76
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	77 def judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(node_name,Final_list_passed):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	78 #used to predict contig is fliC or fljB, if the differnce score value on above head_and_tail is less than 10 (quite small)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	79 #also used when no head or tail got blasted score for the contig
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	80 role=""
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	81 for z in Final_list_passed:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	82 if node_name in z[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	83 role=z[0].split("_")[0]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	84 break
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	85 return role
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	86
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	87
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	88 def fliC_or_fljB_judge_from_head_tail_sequence(nodes_list,tail_head_list,Final_list_passed):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	89 #nodes_list is the c created by c,d=Uniq(nodes) in below function
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	90 first_target=""
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	91 role_list=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	92 for x in nodes_list:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	93 a=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	94 role=""
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	95 for y in tail_head_list:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	96 if x in y[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	97 a.append(y)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	98 if len(a)==4:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	99 #compare two heads (37 > 30)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	100 #four contigs, most perfect assembly, high quality
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	101 """
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	102 for z in a:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	103 if "fliC_first_37" in z[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	104 t1=z[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	105 elif "fljB_first_37" in z[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	106 t2=z[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	107 if t1>=t2:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	108 role="fliC"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	109 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	110 role="fljB"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	111 """
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	112 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	113 if diff<20:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	114 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	115 elif len(a)==3:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	116 """
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	117 #compare the number, because hybrid problem
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	118 temp=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	119 for z in a:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	120 temp.append(z[0].split("_")[0])
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	121 m,n=Uniq(temp)#only two choices in m or n
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	122 if n[0]>n[1]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	123 role=m[0]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	124 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	125 role=m[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	126 """
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	127 ###however, if the one with highest score is the fewer one, compare their accumulation score
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	128 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	129 if diff<20:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	130 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	131 ###end of above score comparison
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	132 elif len(a)==2:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	133 #must on same node, if not, then decide with unit blast score, blast-score/length_of_special_sequence(30 or 37)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	134 temp=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	135 for z in a:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	136 temp.append(z[0].split("_")[0])
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	137 m,n=Uniq(temp)#should only have one choice, but weird situation might occur too
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	138 if len(m)==1:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	139 pass
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	140 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	141 pass
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	142 #print "head and tail not belong to same role, now let's guess based on maximum likelihood"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	143 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	144 if diff<20:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	145 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	146 """
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	147 max_unit_score=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	148 for z in a:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	149 unit_score=z[-1]/int(z[0].split("__")[1])
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	150 if unit_score>=max_unit_score:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	151 role=z[0].split("_")[0]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	152 max_unit_score=unit_score
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	153 """
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	154 ###need to desgin a algorithm to guess most possible situation for nodes_list, See the situations of test evaluation
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	155 elif len(a)==1:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	156 #that one
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	157 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	158 if diff<20:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	159 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	160 #role=a[0][0].split("_")[0]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	161 #need to evaluate, in future, may set up a cut-off, if not met, then just find Final_list_passed best match,like when "a==0"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	162 else:#a==0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	163 #use Final_list_passed best match
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	164 for z in Final_list_passed:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	165 if x in z[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	166 role=z[0].split("_")[0]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	167 break
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	168 #print x,role,len(a)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	169 role_list.append((role,x))
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	170 if len(role_list)==2:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	171 if role_list[0][0]==role_list[1][0]:#this is the most cocmmon error, two antigen were assigned to same phase
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	172 #just use score to do a final test
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	173 role_list=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	174 for x in nodes_list:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	175 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	176 role_list.append((role,x))
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	177 return role_list
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	178
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	179 def decide_contig_roles_for_H_antigen(Final_list):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	180 #used to decide which contig is FliC and which one is fljB
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	181 contigs=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	182 Final_list_passed=[x for x in Final_list if float(x[0].split("_cov_")[1])>=3.5 and (x[1]>=int(x[0].split("__")[1]) or x[1]>=int(x[0].split("___")[1].split("_")[3]))]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	183 nodes=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	184 for x in Final_list_passed:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	185 if x[0].startswith("fl") and "last" not in x[0] and "first" not in x[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	186 nodes.append(x[0].split("___")[1].strip())
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	187 c,d=Uniq(nodes)#c is node_list
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	188 #print c
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	189 tail_head_list=[x for x in Final_list if ("last" in x[0] or "first" in x[0])]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	190 roles=fliC_or_fljB_judge_from_head_tail_sequence(c,tail_head_list,Final_list_passed)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	191 return roles
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	192
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	193 def Combine(b,c):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	194 fliC_combinations=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	195 fliC_combinations.append(",".join(c))
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	196 temp_combinations=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	197 for i in range(len(b)):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	198 for x in itertools.combinations(b,i+1):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	199 temp_combinations.append(",".join(x))
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	200 for x in temp_combinations:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	201 temp=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	202 for y in c:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	203 temp.append(y)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	204 temp.append(x)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	205 temp=",".join(temp)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	206 temp=temp.split(",")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	207 temp.sort()
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	208 temp=",".join(temp)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	209 fliC_combinations.append(temp)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	210 return fliC_combinations
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	211
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	212 def decide_O_type_and_get_special_genes(Final_list):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	213 #decide O based on Final_list
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	214 O_choice="?"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	215 O_list=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	216 special_genes=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	217 Final_list_passed=[x for x in Final_list if float(x[0].split("_cov_")[1])>=3.5 and (x[1]>=int(x[0].split("__")[1]) or x[1]>=int(x[0].split("___")[1].split("_")[3]))]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	218 nodes=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	219 for x in Final_list_passed:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	220 if x[0].startswith("O-"):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	221 nodes.append(x[0].split("___")[1].strip())
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	222 elif not x[0].startswith("fl"):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	223 special_genes.append(x)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	224 #print "special_genes:",special_genes
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	225 c,d=Uniq(nodes)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	226 #print "potential O antigen contig",c
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	227 final_O=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	228 O_nodes_list=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	229 for x in c:#c is the list for contigs
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	230 temp=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	231 for y in Final_list_passed:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	232 if x in y[0] and y[0].startswith("O-"):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	233 final_O.append(y)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	234 break
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	235 ### O contig has the problem of two genes on same contig, so do additional test
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	236 potenial_new_gene=""
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	237 for x in final_O:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	238 pointer=0 #for genes merged or not
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	239 #not consider O-1,3,19_not_in_3,10, too short compared with others
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	240 if "O-1,3,19_not_in_3,10" not in x[0] and int(x[0].split("__")[1].split("___")[0])+800 <= int(x[0].split("length_")[1].split("_")[0]):#gene length << contig length; for now give 3002 (for secureity can use 4002) as flank region
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	241 pointer=x[0].split("___")[1].strip()#store the contig name
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	242 print pointer
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	243 if pointer!=0:#it has potential merge event
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	244 for y in Final_list:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	245 if pointer in y[0] and y not in final_O and (y[1]>=int(y[0].split("__")[1].split("___")[0])1.5 or (y[1]>=int(y[0].split("__")[1].split("___")[0])y[2] and y[1]>=400)):#that's a realtively strict filter now; if passed, it has merge event and add one more to final_O
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	246 potenial_new_gene=y
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	247 print potenial_new_gene
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	248 break
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	249 if potenial_new_gene!="":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	250 print "two differnt genes in same contig, fix it for O antigen"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	251 final_O.append(potenial_new_gene)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	252 ### end of the two genes on same contig test
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	253 if len(final_O)==0:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	254 #print "$$$No Otype, due to no hit"#may need to be changed
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	255 O_choice="-"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	256 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	257 O_list=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	258 for x in final_O:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	259 O_list.append(x[0].split("__")[0])
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	260 if not "O-1,3,19_not_in_3,10__130" in x[0]:#O-1,3,19_not_in_3,10 is too small, which may affect further analysis
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	261 O_nodes_list.append(x[0].split("___")[1])
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	262 ### special test for O9,46 and O3,10 family
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	263 if "O-9,46_wbaV" in O_list:#not sure should use and float(O9_wbaV)/float(num_1) > 0.1
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	264 if "O-9,46_wzy" in O_list:#and float(O946_wzy)/float(num_1) > 0.1
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	265 O_choice="O-9,46"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	266 #print "$$$Most possilble Otype: O-9,46"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	267 elif "O-9,46,27_partial_wzy" in O_list:#and float(O94627)/float(num_1) > 0.1
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	268 O_choice="O-9,46,27"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	269 #print "$$$Most possilble Otype: O-9,46,27"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	270 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	271 O_choice="O-9"#next, detect O9 vs O2?
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	272 O2=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	273 O9=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	274 for z in special_genes:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	275 if "tyr-O-9" in z[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	276 O9=z[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	277 elif "tyr-O-2" in z[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	278 O2=z[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	279 if O2>O9:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	280 O_choice="O-2"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	281 elif O2<O9:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	282 pass
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	283 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	284 pass
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	285 #print "$$$No suitable one, because can't distinct it's O-9 or O-2, but O-9 has a more possibility."
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	286 elif ("O-3,10_wzx" in O_list) and ("O-9,46_wzy" in O_list):#and float(O310_wzx)/float(num_1) > 0.1 and float(O946_wzy)/float(num_1) > 0.1
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	287 if "O-3,10_not_in_1,3,19" in O_list:#and float(O310_no_1319)/float(num_1) > 0.1
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	288 O_choice="O-3,10"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	289 #print "$$$Most possilble Otype: O-3,10 (contain O-3,10_not_in_1,3,19)"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	290 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	291 O_choice="O-1,3,19"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	292 #print "$$$Most possilble Otype: O-1,3,19 (not contain O-3,10_not_in_1,3,19)"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	293 ### end of special test for O9,46 and O3,10 family
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	294 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	295 try:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	296 max_score=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	297 for x in final_O:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	298 if x[1]>=max_score:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	299 max_score=x[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	300 O_choice=x[0].split("_")[0]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	301 if O_choice=="O-1,3,19":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	302 O_choice=final_O[1][0].split("_")[0]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	303 #print "$$$Most possilble Otype: ",O_choice
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	304 except:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	305 pass
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	306 #print "$$$No suitable Otype, or failure of mapping (please check the quality of raw reads)"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	307 #print "O:",O_choice,O_nodes_list
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	308 return O_choice,O_nodes_list,special_genes,final_O
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	309
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	310 def seqsero_from_formula_to_serotypes(Otype,fliC,fljB,special_gene_list):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	311 #like test_output_06012017.txt
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	312 #can add more varialbles like sdf-type, sub-species-type in future (we can conclude it into a special-gene-list)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	313 from Initial_Conditions import phase1
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	314 from Initial_Conditions import phase2
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	315 from Initial_Conditions import phaseO
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	316 from Initial_Conditions import sero
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	317 seronames=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	318 for i in range(len(phase1)):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	319 fliC_combine=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	320 fljB_combine=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	321 if phaseO[i]==Otype:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	322 ### for fliC, detect every possible combinations to avoid the effect of "["
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	323 if phase1[i].count("[")==0:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	324 fliC_combine.append(phase1[i])
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	325 elif phase1[i].count("[")>=1:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	326 c=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	327 b=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	328 if phase1[i][0]=="[" and phase1[i][-1]=="]" and phase1[i].count("[")==1:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	329 content=phase1[i].replace("[","").replace("]","")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	330 fliC_combine.append(content)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	331 fliC_combine.append("-")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	332 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	333 for x in phase1[i].split(","):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	334 if "[" in x:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	335 b.append(x.replace("[","").replace("]",""))
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	336 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	337 c.append(x)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	338 fliC_combine=Combine(b,c) #Combine will offer every possible combinations of the formula, like f,[g],t: f,t f,g,t
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	339 ### end of fliC "[" detect
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	340 ### for fljB, detect every possible combinations to avoid the effect of "["
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	341 if phase2[i].count("[")==0:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	342 fljB_combine.append(phase2[i])
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	343 elif phase2[i].count("[")>=1:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	344 d=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	345 e=[]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	346 if phase2[i][0]=="[" and phase2[i][-1]=="]" and phase2[i].count("[")==1:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	347 content=phase2[i].replace("[","").replace("]","")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	348 fljB_combine.append(content)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	349 fljB_combine.append("-")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	350 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	351 for x in phase2[i].split(","):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	352 if "[" in x:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	353 d.append(x.replace("[","").replace("]",""))
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	354 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	355 e.append(x)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	356 fljB_combine=Combine(d,e)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	357 ### end of fljB "[" detect
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	358 new_fliC=fliC.split(",") #because some antigen like r,[i] not follow alphabetical order, so use this one to judge and can avoid missings
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	359 new_fliC.sort()
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	360 new_fliC=",".join(new_fliC)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	361 new_fljB=fljB.split(",")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	362 new_fljB.sort()
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	363 new_fljB=",".join(new_fljB)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	364 if (new_fliC in fliC_combine or fliC in fliC_combine) and (new_fljB in fljB_combine or fljB in fljB_combine):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	365 seronames.append(sero[i])
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	366 #analyze seronames
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	367 if len(seronames)==0:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	368 seronames=["N/A (The predicted antigenic profile does not exist in the White-Kauffmann-Le Minor scheme)"]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	369 star=""
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	370 star_line=""
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	371 if len(seronames)>1:#there are two possible predictions for serotypes
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	372 star="*"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	373 star_line="The predicted serotypes share the same general formula:\t"+Otype+":"+fliC+":"+fljB+"\n"##
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	374 print "\n"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	375 predict_form=Otype+":"+fliC+":"+fljB#
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	376 predict_sero=(" or ").join(seronames)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	377 ###special test for Enteritidis
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	378 if predict_form=="9:g,m:-":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	379 sdf="-"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	380 for x in special_gene_list:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	381 if x[0].startswith("sdf"):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	382 sdf="+"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	383 predict_form=predict_form+"\nSdf prediction:"+sdf
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	384 if sdf=="-":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	385 star="*"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	386 star_line="Additional characterization is necessary to assign a serotype to this strain. Commonly circulating strains of serotype Enteritidis are sdf+, although sdf- strains of serotype Enteritidis are known to exist. Serotype Gallinarum is typically sdf- but should be quite rare. Sdf- strains of serotype Enteritidis and serotype Gallinarum can be differentiated by phenotypic profile or genetic criteria.\n"#+##
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	387 predict_sero="See comments below"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	388 ###end of special test for Enteritidis
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	389 elif predict_form=="4:i:-":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	390 predict_sero="potential monophasic variant of Typhimurium"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	391 elif predict_form=="4:r:-":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	392 predict_sero="potential monophasic variant of Heidelberg"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	393 elif predict_form=="4:b:-":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	394 predict_sero="potential monophasic variant of Paratyphi B"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	395 elif predict_form=="8:e,h:1,2":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	396 predict_sero="Newport"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	397 star="*"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	398 star_line="Serotype Bardo shares the same antigenic profile with Newport, but Bardo is exceedingly rare."
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	399 claim="The serotype(s) is/are the only serotype(s) with the indicated antigenic profile currently recognized in the Kauffmann White Scheme. New serotypes can emerge and the possibility exists that this antigenic profile may emerge in a different subspecies. Identification of strains to the subspecies level should accompany serotype determination; the same antigenic profile in different subspecies is considered different serotypes."##
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	400 if "N/A" in predict_sero:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	401 claim=""
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	402 if "Typhimurium" in predict_sero or predict_form=="4:i:-":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	403 normal=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	404 mutation=0
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	405 for x in special_gene_list:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	406 if "oafA-O-4_full" in x[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	407 normal=x[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	408 elif "oafA-O-4_5-" in x[0]:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	409 mutation=x[1]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	410 if normal>mutation:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	411 #print "$$$Typhimurium"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	412 pass
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	413 elif normal<mutation:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	414 predict_sero=predict_sero.strip()+"(O5-)"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	415 star="*"#
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	416 star_line="Detected the deletion of O5-."
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	417 #print "$$$Typhimurium_O5-"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	418 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	419 #print "$$$Typhimurium, even no 7 bases difference"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	420 pass
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	421 return predict_form,predict_sero,star,star_line,claim
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	422
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	423 def main():
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	424 database=sys.argv[1]#used to extract reads
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	425 mapping_mode=sys.argv[2]#mem or sampe
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	426 threads=sys.argv[3]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	427 for_fq=sys.argv[4]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	428 rev_fq=sys.argv[5]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	429 current_time=time.strftime("%Y_%m_%d_%H_%M_%S", time.localtime())
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	430 sam=for_fq+".sam"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	431 bam=for_fq+".bam"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	432 sorted_bam=for_fq+"_sorted.bam"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	433 mapped_fq1=for_fq+"_mapped.fq"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	434 mapped_fq2=rev_fq+"_mapped.fq"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	435 combined_fq=for_fq+"_combined.fq"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	436 for_sai=for_fq+".sai"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	437 rev_sai=rev_fq+".sai"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	438 print "building database..."
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	439 #os.system("bwa index "+database+ " 2> /dev/null")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	440 os.system("bwa index "+database+ " 2>> data_log.txt ")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	441 print "mapping..."
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	442 if mapping_mode=="mem":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	443 os.system("bwa mem -t "+threads+" "+database+" "+for_fq+" "+rev_fq+" > "+sam+ " 2>> data_log.txt")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	444 elif mapping_mode=="sam":
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	445 os.system("bwa aln -t "+threads+" "+database+" "+for_fq+" > "+for_sai+ " 2>> data_log.txt")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	446 os.system("bwa aln -t "+threads+" "+database+" "+rev_fq+" > "+rev_sai+ " 2>> data_log.txt")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	447 os.system("bwa sampe "+database+" "+for_sai+" "+ rev_sai+" "+for_fq+" "+rev_fq+" > "+sam+ " 2>> data_log.txt")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	448 os.system("samtools view -@ "+threads+" -F 4 -Sbh "+sam+" > "+bam)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	449 os.system("samtools view -@ "+threads+" -h -o "+sam+" "+bam)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	450 ### check the version of samtools then use differnt commands
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	451 samtools_version=subprocess.Popen(["samtools"],stdout=subprocess.PIPE,stderr=subprocess.PIPE)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	452 out, err = samtools_version.communicate()
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	453 version = err.split("ersion:")[1].strip().split(" ")[0].strip()
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	454 print "check samtools version:",version
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	455 if LooseVersion(version)<=LooseVersion("1.2"):
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	456 os.system("samtools sort -@ "+threads+" -n "+bam+" "+for_fq+"_sorted")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	457 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	458 os.system("samtools sort -@ "+threads+" -n "+bam+" >"+sorted_bam)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	459 ### end of samtools version check and its analysis
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	460 os.system("bamToFastq -i "+sorted_bam+" -fq "+combined_fq)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	461 os.system("bamToFastq -i "+sorted_bam+" -fq "+mapped_fq1+" -fq2 "+mapped_fq2 + " 2>> data_log.txt")#2> /dev/null if want no output
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	462 outdir=current_time+"_temp"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	463 print "assembling..."
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	464 if int(threads)>4:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	465 t="4"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	466 else:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	467 t=threads
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	468 os.system("spades.py --careful --pe1-s "+combined_fq+" --pe1-1 "+mapped_fq1+" --pe1-2 "+mapped_fq2+" -t "+t+" -o "+outdir+ " >> data_log.txt 2>&1")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	469 new_fasta=for_fq+"_"+database+"_"+mapping_mode+".fasta"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	470 os.system("mv "+outdir+"/contigs.fasta "+new_fasta+ " 2> /dev/null")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	471 #os.system("mv "+outdir+"/scaffolds.fasta "+new_fasta+ " 2> /dev/null") contigs.fasta
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	472 os.system("rm -rf "+outdir+ " 2> /dev/null")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	473 ### begin blast
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	474 print "blasting..."
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	475 print "\n"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	476 xmlfile=for_fq+"-extracted_vs_"+database+"_"+mapping_mode+".xml"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	477 os.system('makeblastdb -in '+new_fasta+' -out '+new_fasta+'_db '+'-dbtype nucl >> data_log.txt 2>&1') #temp.txt is to forbid the blast result interrupt the output of our program###1/27/2015
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	478 os.system("blastn -word_size 10 -query "+database+" -db "+new_fasta+"_db -out "+xmlfile+" -outfmt 5 >> data_log.txt 2>&1")###1/27/2015
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	479 Final_list=xml_parse_score_comparision_seqsero(xmlfile)
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	480 Final_list_passed=[x for x in Final_list if float(x[0].split("_cov_")[1])>=3.5 and (x[1]>=int(x[0].split("__")[1]) or x[1]>=int(x[0].split("___")[1].split("_")[3]))]
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	481 fliC_choice="-"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	482 fljB_choice="-"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	483 fliC_contig="NA"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	484 fljB_contig="NA"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	485 fliC_length=0 #can be changed to coverage in future
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	486 fljB_length=0 #can be changed to coverage in future
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	487 O_choice=""#no need to decide O contig for now, should be only one
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	488 O_choice,O_nodes,special_gene_list,O_nodes_roles=decide_O_type_and_get_special_genes(Final_list)#decide the O antigen type and also return special-gene-list for further identification
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	489 O_choice=O_choice.split("-")[-1].strip()
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	490 H_contig_roles=decide_contig_roles_for_H_antigen(Final_list)#decide the H antigen contig is fliC or fljB
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	491 log_file=open("SeqSero_hybrid_assembly_log.txt","a")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	492 print "O_contigs:"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	493 log_file.write("O_contigs:\n")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	494 for x in O_nodes_roles:
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	495 if "O-1,3,19_not_in_3,10" not in x[0]:#O-1,3,19_not_in_3,10 is just a small size marker
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	496 print x[0].split("___")[-1],x[0].split("__")[0],"blast score:",x[1],"identity%:",str(round(x[2]*100,2))+"%"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	497 log_file.write(x[0].split("___")[-1]+" "+x[0].split("__")[0]+" "+"blast score: "+str(x[1])+"identity%:"+str(round(x[2]*100,2))+"%"+"\n")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	498 print "H_contigs:"
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	499 log_file.write("H_contigs:\n")
6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty cstrittmatter parents: diff changeset	500 H_contig_stat=[]

0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

1 import os,sys,glob,time,itertools,subprocess

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

2 from Initial_Conditions import phase1

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

3 from Initial_Conditions import phase2

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

4 from Initial_Conditions import phaseO

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

5 from Initial_Conditions import sero

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

6 from distutils.version import LooseVersion

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

7

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

8

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

9

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

10

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

11 def xml_parse_score_comparision_seqsero(xmlfile):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

12 #used to do seqsero xml analysis

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

13 from Bio.Blast import NCBIXML

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

14 handle=open(xmlfile)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

15 handle=NCBIXML.parse(handle)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

16 handle=list(handle)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

17 List=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

18 List_score=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

19 List_ids=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

20 for i in range(len(handle)):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

21 if len(handle[i].alignments)>0:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

22 for j in range(len(handle[i].alignments)):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

23 score=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

24 ids=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

25 List.append(handle[i].query.strip()+"___"+handle[i].alignments[j].hit_def)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

26 for z in range(len(handle[i].alignments[j].hsps)):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

27 if "last" in handle[i].query or "first" in handle[i].query:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

28 score+=handle[i].alignments[j].hsps[z].bits

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

29 ids+=float(handle[i].alignments[j].hsps[z].identities)/handle[i].query_length

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

30 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

31 if handle[i].alignments[j].hsps[z].align_length>=30:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

32 #for the long alleles, filter noise parts

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

33 score+=handle[i].alignments[j].hsps[z].bits

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

34 ids+=float(handle[i].alignments[j].hsps[z].identities)/handle[i].query_length

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

35 List_score.append(score)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

36 List_ids.append(ids)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

37 temp=zip(List,List_score,List_ids)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

38 Final_list=sorted(temp, key=lambda d:d[1], reverse = True)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

39 return Final_list

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

40

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

41

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

42 def Uniq(L,sort_on_fre="none"): #return the uniq list and the count number

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

43 Old=L

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

44 L.sort()

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

45 L = [L[i] for i in range(len(L)) if L[i] not in L[:i]]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

46 count=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

47 for j in range(len(L)):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

48 y=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

49 for x in Old:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

50 if L[j]==x:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

51 y+=1

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

52 count.append(y)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

53 if sort_on_fre!="none":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

54 d=zip(*sorted(zip(count, L)))

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

55 L=d[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

56 count=d[0]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

57 return (L,count)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

58

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

59

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

60 def judge_fliC_or_fljB_from_head_tail_for_one_contig(nodes_vs_score_list):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

61 #used to predict it's fliC or fljB for one contig, based on tail and head score, but output the score difference,if it is very small, then not reliable, use blast score for whole contig to test

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

62 #this is mainly used for

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

63 a=nodes_vs_score_list

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

64 fliC_score=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

65 fljB_score=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

66 for z in a:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

67 if "fliC" in z[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

68 fliC_score+=z[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

69 elif "fljB" in z[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

70 fljB_score+=z[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

71 if fliC_score>=fljB_score:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

72 role="fliC"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

73 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

74 role="fljB"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

75 return (role,abs(fliC_score-fljB_score))

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

76

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

77 def judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(node_name,Final_list_passed):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

78 #used to predict contig is fliC or fljB, if the differnce score value on above head_and_tail is less than 10 (quite small)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

79 #also used when no head or tail got blasted score for the contig

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

80 role=""

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

81 for z in Final_list_passed:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

82 if node_name in z[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

83 role=z[0].split("_")[0]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

84 break

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

85 return role

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

86

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

87

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

88 def fliC_or_fljB_judge_from_head_tail_sequence(nodes_list,tail_head_list,Final_list_passed):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

89 #nodes_list is the c created by c,d=Uniq(nodes) in below function

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

90 first_target=""

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

91 role_list=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

92 for x in nodes_list:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

93 a=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

94 role=""

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

95 for y in tail_head_list:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

96 if x in y[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

97 a.append(y)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

98 if len(a)==4:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

99 #compare two heads (37 > 30)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

100 #four contigs, most perfect assembly, high quality

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

101 """

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

102 for z in a:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

103 if "fliC_first_37" in z[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

104 t1=z[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

105 elif "fljB_first_37" in z[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

106 t2=z[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

107 if t1>=t2:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

108 role="fliC"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

109 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

110 role="fljB"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

111 """

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

112 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

113 if diff<20:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

114 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

115 elif len(a)==3:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

116 """

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

117 #compare the number, because hybrid problem

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

118 temp=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

119 for z in a:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

120 temp.append(z[0].split("_")[0])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

121 m,n=Uniq(temp)#only two choices in m or n

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

122 if n[0]>n[1]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

123 role=m[0]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

124 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

125 role=m[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

126 """

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

127 ###however, if the one with highest score is the fewer one, compare their accumulation score

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

128 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

129 if diff<20:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

130 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

131 ###end of above score comparison

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

132 elif len(a)==2:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

133 #must on same node, if not, then decide with unit blast score, blast-score/length_of_special_sequence(30 or 37)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

134 temp=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

135 for z in a:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

136 temp.append(z[0].split("_")[0])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

137 m,n=Uniq(temp)#should only have one choice, but weird situation might occur too

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

138 if len(m)==1:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

139 pass

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

140 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

141 pass

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

142 #print "head and tail not belong to same role, now let's guess based on maximum likelihood"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

143 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

144 if diff<20:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

145 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

146 """

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

147 max_unit_score=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

148 for z in a:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

149 unit_score=z[-1]/int(z[0].split("__")[1])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

150 if unit_score>=max_unit_score:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

151 role=z[0].split("_")[0]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

152 max_unit_score=unit_score

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

153 """

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

154 ###need to desgin a algorithm to guess most possible situation for nodes_list, See the situations of test evaluation

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

155 elif len(a)==1:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

156 #that one

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

157 role,diff=judge_fliC_or_fljB_from_head_tail_for_one_contig(a)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

158 if diff<20:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

159 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

160 #role=a[0][0].split("_")[0]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

161 #need to evaluate, in future, may set up a cut-off, if not met, then just find Final_list_passed best match,like when "a==0"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

162 else:#a==0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

163 #use Final_list_passed best match

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

164 for z in Final_list_passed:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

165 if x in z[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

166 role=z[0].split("_")[0]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

167 break

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

168 #print x,role,len(a)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

169 role_list.append((role,x))

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

170 if len(role_list)==2:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

171 if role_list[0][0]==role_list[1][0]:#this is the most cocmmon error, two antigen were assigned to same phase

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

172 #just use score to do a final test

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

173 role_list=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

174 for x in nodes_list:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

175 role=judge_fliC_or_fljB_from_whole_contig_blast_score_ranking(x,Final_list_passed)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

176 role_list.append((role,x))

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

177 return role_list

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

178

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

179 def decide_contig_roles_for_H_antigen(Final_list):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

180 #used to decide which contig is FliC and which one is fljB

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

181 contigs=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

182 Final_list_passed=[x for x in Final_list if float(x[0].split("_cov_")[1])>=3.5 and (x[1]>=int(x[0].split("__")[1]) or x[1]>=int(x[0].split("___")[1].split("_")[3]))]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

183 nodes=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

184 for x in Final_list_passed:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

185 if x[0].startswith("fl") and "last" not in x[0] and "first" not in x[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

186 nodes.append(x[0].split("___")[1].strip())

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

187 c,d=Uniq(nodes)#c is node_list

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

188 #print c

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

189 tail_head_list=[x for x in Final_list if ("last" in x[0] or "first" in x[0])]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

190 roles=fliC_or_fljB_judge_from_head_tail_sequence(c,tail_head_list,Final_list_passed)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

191 return roles

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

192

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

193 def Combine(b,c):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

194 fliC_combinations=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

195 fliC_combinations.append(",".join(c))

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

196 temp_combinations=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

197 for i in range(len(b)):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

198 for x in itertools.combinations(b,i+1):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

199 temp_combinations.append(",".join(x))

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

200 for x in temp_combinations:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

201 temp=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

202 for y in c:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

203 temp.append(y)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

204 temp.append(x)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

205 temp=",".join(temp)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

206 temp=temp.split(",")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

207 temp.sort()

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

208 temp=",".join(temp)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

209 fliC_combinations.append(temp)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

210 return fliC_combinations

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

211

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

212 def decide_O_type_and_get_special_genes(Final_list):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

213 #decide O based on Final_list

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

214 O_choice="?"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

215 O_list=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

216 special_genes=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

217 Final_list_passed=[x for x in Final_list if float(x[0].split("_cov_")[1])>=3.5 and (x[1]>=int(x[0].split("__")[1]) or x[1]>=int(x[0].split("___")[1].split("_")[3]))]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

218 nodes=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

219 for x in Final_list_passed:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

220 if x[0].startswith("O-"):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

221 nodes.append(x[0].split("___")[1].strip())

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

222 elif not x[0].startswith("fl"):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

223 special_genes.append(x)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

224 #print "special_genes:",special_genes

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

225 c,d=Uniq(nodes)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

226 #print "potential O antigen contig",c

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

227 final_O=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

228 O_nodes_list=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

229 for x in c:#c is the list for contigs

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

230 temp=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

231 for y in Final_list_passed:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

232 if x in y[0] and y[0].startswith("O-"):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

233 final_O.append(y)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

234 break

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

235 ### O contig has the problem of two genes on same contig, so do additional test

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

236 potenial_new_gene=""

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

237 for x in final_O:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

238 pointer=0 #for genes merged or not

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

239 #not consider O-1,3,19_not_in_3,10, too short compared with others

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

240 if "O-1,3,19_not_in_3,10" not in x[0] and int(x[0].split("__")[1].split("___")[0])+800 <= int(x[0].split("length_")[1].split("_")[0]):#gene length << contig length; for now give 300*2 (for secureity can use 400*2) as flank region

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

241 pointer=x[0].split("___")[1].strip()#store the contig name

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

242 print pointer

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

243 if pointer!=0:#it has potential merge event

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

244 for y in Final_list:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

245 if pointer in y[0] and y not in final_O and (y[1]>=int(y[0].split("__")[1].split("___")[0])*1.5 or (y[1]>=int(y[0].split("__")[1].split("___")[0])*y[2] and y[1]>=400)):#that's a realtively strict filter now; if passed, it has merge event and add one more to final_O

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

246 potenial_new_gene=y

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

247 print potenial_new_gene

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

248 break

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

249 if potenial_new_gene!="":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

250 print "two differnt genes in same contig, fix it for O antigen"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

251 final_O.append(potenial_new_gene)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

252 ### end of the two genes on same contig test

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

253 if len(final_O)==0:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

254 #print "$$$No Otype, due to no hit"#may need to be changed

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

255 O_choice="-"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

256 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

257 O_list=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

258 for x in final_O:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

259 O_list.append(x[0].split("__")[0])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

260 if not "O-1,3,19_not_in_3,10__130" in x[0]:#O-1,3,19_not_in_3,10 is too small, which may affect further analysis

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

261 O_nodes_list.append(x[0].split("___")[1])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

262 ### special test for O9,46 and O3,10 family

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

263 if "O-9,46_wbaV" in O_list:#not sure should use and float(O9_wbaV)/float(num_1) > 0.1

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

264 if "O-9,46_wzy" in O_list:#and float(O946_wzy)/float(num_1) > 0.1

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

265 O_choice="O-9,46"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

266 #print "$$$Most possilble Otype: O-9,46"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

267 elif "O-9,46,27_partial_wzy" in O_list:#and float(O94627)/float(num_1) > 0.1

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

268 O_choice="O-9,46,27"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

269 #print "$$$Most possilble Otype: O-9,46,27"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

270 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

271 O_choice="O-9"#next, detect O9 vs O2?

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

272 O2=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

273 O9=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

274 for z in special_genes:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

275 if "tyr-O-9" in z[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

276 O9=z[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

277 elif "tyr-O-2" in z[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

278 O2=z[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

279 if O2>O9:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

280 O_choice="O-2"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

281 elif O2<O9:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

282 pass

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

283 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

284 pass

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

285 #print "$$$No suitable one, because can't distinct it's O-9 or O-2, but O-9 has a more possibility."

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

286 elif ("O-3,10_wzx" in O_list) and ("O-9,46_wzy" in O_list):#and float(O310_wzx)/float(num_1) > 0.1 and float(O946_wzy)/float(num_1) > 0.1

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

287 if "O-3,10_not_in_1,3,19" in O_list:#and float(O310_no_1319)/float(num_1) > 0.1

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

288 O_choice="O-3,10"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

289 #print "$$$Most possilble Otype: O-3,10 (contain O-3,10_not_in_1,3,19)"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

290 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

291 O_choice="O-1,3,19"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

292 #print "$$$Most possilble Otype: O-1,3,19 (not contain O-3,10_not_in_1,3,19)"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

293 ### end of special test for O9,46 and O3,10 family

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

294 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

295 try:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

296 max_score=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

297 for x in final_O:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

298 if x[1]>=max_score:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

299 max_score=x[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

300 O_choice=x[0].split("_")[0]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

301 if O_choice=="O-1,3,19":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

302 O_choice=final_O[1][0].split("_")[0]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

303 #print "$$$Most possilble Otype: ",O_choice

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

304 except:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

305 pass

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

306 #print "$$$No suitable Otype, or failure of mapping (please check the quality of raw reads)"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

307 #print "O:",O_choice,O_nodes_list

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

308 return O_choice,O_nodes_list,special_genes,final_O

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

309

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

310 def seqsero_from_formula_to_serotypes(Otype,fliC,fljB,special_gene_list):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

311 #like test_output_06012017.txt

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

312 #can add more varialbles like sdf-type, sub-species-type in future (we can conclude it into a special-gene-list)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

313 from Initial_Conditions import phase1

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

314 from Initial_Conditions import phase2

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

315 from Initial_Conditions import phaseO

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

316 from Initial_Conditions import sero

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

317 seronames=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

318 for i in range(len(phase1)):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

319 fliC_combine=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

320 fljB_combine=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

321 if phaseO[i]==Otype:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

322 ### for fliC, detect every possible combinations to avoid the effect of "["

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

323 if phase1[i].count("[")==0:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

324 fliC_combine.append(phase1[i])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

325 elif phase1[i].count("[")>=1:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

326 c=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

327 b=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

328 if phase1[i][0]=="[" and phase1[i][-1]=="]" and phase1[i].count("[")==1:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

329 content=phase1[i].replace("[","").replace("]","")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

330 fliC_combine.append(content)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

331 fliC_combine.append("-")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

332 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

333 for x in phase1[i].split(","):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

334 if "[" in x:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

335 b.append(x.replace("[","").replace("]",""))

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

336 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

337 c.append(x)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

338 fliC_combine=Combine(b,c) #Combine will offer every possible combinations of the formula, like f,[g],t: f,t f,g,t

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

339 ### end of fliC "[" detect

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

340 ### for fljB, detect every possible combinations to avoid the effect of "["

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

341 if phase2[i].count("[")==0:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

342 fljB_combine.append(phase2[i])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

343 elif phase2[i].count("[")>=1:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

344 d=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

345 e=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

346 if phase2[i][0]=="[" and phase2[i][-1]=="]" and phase2[i].count("[")==1:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

347 content=phase2[i].replace("[","").replace("]","")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

348 fljB_combine.append(content)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

349 fljB_combine.append("-")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

350 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

351 for x in phase2[i].split(","):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

352 if "[" in x:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

353 d.append(x.replace("[","").replace("]",""))

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

354 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

355 e.append(x)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

356 fljB_combine=Combine(d,e)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

357 ### end of fljB "[" detect

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

358 new_fliC=fliC.split(",") #because some antigen like r,[i] not follow alphabetical order, so use this one to judge and can avoid missings

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

359 new_fliC.sort()

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

360 new_fliC=",".join(new_fliC)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

361 new_fljB=fljB.split(",")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

362 new_fljB.sort()

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

363 new_fljB=",".join(new_fljB)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

364 if (new_fliC in fliC_combine or fliC in fliC_combine) and (new_fljB in fljB_combine or fljB in fljB_combine):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

365 seronames.append(sero[i])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

366 #analyze seronames

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

367 if len(seronames)==0:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

368 seronames=["N/A (The predicted antigenic profile does not exist in the White-Kauffmann-Le Minor scheme)"]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

369 star=""

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

370 star_line=""

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

371 if len(seronames)>1:#there are two possible predictions for serotypes

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

372 star="*"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

373 star_line="The predicted serotypes share the same general formula:\t"+Otype+":"+fliC+":"+fljB+"\n"##

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

374 print "\n"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

375 predict_form=Otype+":"+fliC+":"+fljB#

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

376 predict_sero=(" or ").join(seronames)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

377 ###special test for Enteritidis

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

378 if predict_form=="9:g,m:-":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

379 sdf="-"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

380 for x in special_gene_list:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

381 if x[0].startswith("sdf"):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

382 sdf="+"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

383 predict_form=predict_form+"\nSdf prediction:"+sdf

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

384 if sdf=="-":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

385 star="*"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

386 star_line="Additional characterization is necessary to assign a serotype to this strain. Commonly circulating strains of serotype Enteritidis are sdf+, although sdf- strains of serotype Enteritidis are known to exist. Serotype Gallinarum is typically sdf- but should be quite rare. Sdf- strains of serotype Enteritidis and serotype Gallinarum can be differentiated by phenotypic profile or genetic criteria.\n"#+##

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

387 predict_sero="See comments below"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

388 ###end of special test for Enteritidis

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

389 elif predict_form=="4:i:-":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

390 predict_sero="potential monophasic variant of Typhimurium"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

391 elif predict_form=="4:r:-":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

392 predict_sero="potential monophasic variant of Heidelberg"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

393 elif predict_form=="4:b:-":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

394 predict_sero="potential monophasic variant of Paratyphi B"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

395 elif predict_form=="8:e,h:1,2":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

396 predict_sero="Newport"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

397 star="*"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

398 star_line="Serotype Bardo shares the same antigenic profile with Newport, but Bardo is exceedingly rare."

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

399 claim="The serotype(s) is/are the only serotype(s) with the indicated antigenic profile currently recognized in the Kauffmann White Scheme. New serotypes can emerge and the possibility exists that this antigenic profile may emerge in a different subspecies. Identification of strains to the subspecies level should accompany serotype determination; the same antigenic profile in different subspecies is considered different serotypes."##

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

400 if "N/A" in predict_sero:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

401 claim=""

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

402 if "Typhimurium" in predict_sero or predict_form=="4:i:-":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

403 normal=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

404 mutation=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

405 for x in special_gene_list:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

406 if "oafA-O-4_full" in x[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

407 normal=x[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

408 elif "oafA-O-4_5-" in x[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

409 mutation=x[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

410 if normal>mutation:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

411 #print "$$$Typhimurium"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

412 pass

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

413 elif normal<mutation:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

414 predict_sero=predict_sero.strip()+"(O5-)"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

415 star="*"#

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

416 star_line="Detected the deletion of O5-."

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

417 #print "$$$Typhimurium_O5-"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

418 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

419 #print "$$$Typhimurium, even no 7 bases difference"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

420 pass

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

421 return predict_form,predict_sero,star,star_line,claim

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

422

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

423 def main():

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

424 database=sys.argv[1]#used to extract reads

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

425 mapping_mode=sys.argv[2]#mem or sampe

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

426 threads=sys.argv[3]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

427 for_fq=sys.argv[4]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

428 rev_fq=sys.argv[5]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

429 current_time=time.strftime("%Y_%m_%d_%H_%M_%S", time.localtime())

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

430 sam=for_fq+".sam"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

431 bam=for_fq+".bam"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

432 sorted_bam=for_fq+"_sorted.bam"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

433 mapped_fq1=for_fq+"_mapped.fq"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

434 mapped_fq2=rev_fq+"_mapped.fq"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

435 combined_fq=for_fq+"_combined.fq"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

436 for_sai=for_fq+".sai"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

437 rev_sai=rev_fq+".sai"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

438 print "building database..."

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

439 #os.system("bwa index "+database+ " 2> /dev/null")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

440 os.system("bwa index "+database+ " 2>> data_log.txt ")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

441 print "mapping..."

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

442 if mapping_mode=="mem":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

443 os.system("bwa mem -t "+threads+" "+database+" "+for_fq+" "+rev_fq+" > "+sam+ " 2>> data_log.txt")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

444 elif mapping_mode=="sam":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

445 os.system("bwa aln -t "+threads+" "+database+" "+for_fq+" > "+for_sai+ " 2>> data_log.txt")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

446 os.system("bwa aln -t "+threads+" "+database+" "+rev_fq+" > "+rev_sai+ " 2>> data_log.txt")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

447 os.system("bwa sampe "+database+" "+for_sai+" "+ rev_sai+" "+for_fq+" "+rev_fq+" > "+sam+ " 2>> data_log.txt")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

448 os.system("samtools view -@ "+threads+" -F 4 -Sbh "+sam+" > "+bam)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

449 os.system("samtools view -@ "+threads+" -h -o "+sam+" "+bam)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

450 ### check the version of samtools then use differnt commands

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

451 samtools_version=subprocess.Popen(["samtools"],stdout=subprocess.PIPE,stderr=subprocess.PIPE)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

452 out, err = samtools_version.communicate()

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

453 version = err.split("ersion:")[1].strip().split(" ")[0].strip()

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

454 print "check samtools version:",version

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

455 if LooseVersion(version)<=LooseVersion("1.2"):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

456 os.system("samtools sort -@ "+threads+" -n "+bam+" "+for_fq+"_sorted")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

457 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

458 os.system("samtools sort -@ "+threads+" -n "+bam+" >"+sorted_bam)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

459 ### end of samtools version check and its analysis

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

460 os.system("bamToFastq -i "+sorted_bam+" -fq "+combined_fq)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

461 os.system("bamToFastq -i "+sorted_bam+" -fq "+mapped_fq1+" -fq2 "+mapped_fq2 + " 2>> data_log.txt")#2> /dev/null if want no output

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

462 outdir=current_time+"_temp"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

463 print "assembling..."

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

464 if int(threads)>4:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

465 t="4"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

466 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

467 t=threads

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

468 os.system("spades.py --careful --pe1-s "+combined_fq+" --pe1-1 "+mapped_fq1+" --pe1-2 "+mapped_fq2+" -t "+t+" -o "+outdir+ " >> data_log.txt 2>&1")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

469 new_fasta=for_fq+"_"+database+"_"+mapping_mode+".fasta"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

470 os.system("mv "+outdir+"/contigs.fasta "+new_fasta+ " 2> /dev/null")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

471 #os.system("mv "+outdir+"/scaffolds.fasta "+new_fasta+ " 2> /dev/null") contigs.fasta

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

472 os.system("rm -rf "+outdir+ " 2> /dev/null")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

473 ### begin blast

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

474 print "blasting..."

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

475 print "\n"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

476 xmlfile=for_fq+"-extracted_vs_"+database+"_"+mapping_mode+".xml"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

477 os.system('makeblastdb -in '+new_fasta+' -out '+new_fasta+'_db '+'-dbtype nucl >> data_log.txt 2>&1') #temp.txt is to forbid the blast result interrupt the output of our program###1/27/2015

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

478 os.system("blastn -word_size 10 -query "+database+" -db "+new_fasta+"_db -out "+xmlfile+" -outfmt 5 >> data_log.txt 2>&1")###1/27/2015

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

479 Final_list=xml_parse_score_comparision_seqsero(xmlfile)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

480 Final_list_passed=[x for x in Final_list if float(x[0].split("_cov_")[1])>=3.5 and (x[1]>=int(x[0].split("__")[1]) or x[1]>=int(x[0].split("___")[1].split("_")[3]))]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

481 fliC_choice="-"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

482 fljB_choice="-"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

483 fliC_contig="NA"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

484 fljB_contig="NA"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

485 fliC_length=0 #can be changed to coverage in future

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

486 fljB_length=0 #can be changed to coverage in future

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

487 O_choice=""#no need to decide O contig for now, should be only one

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

488 O_choice,O_nodes,special_gene_list,O_nodes_roles=decide_O_type_and_get_special_genes(Final_list)#decide the O antigen type and also return special-gene-list for further identification

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

489 O_choice=O_choice.split("-")[-1].strip()

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

490 H_contig_roles=decide_contig_roles_for_H_antigen(Final_list)#decide the H antigen contig is fliC or fljB

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

491 log_file=open("SeqSero_hybrid_assembly_log.txt","a")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

492 print "O_contigs:"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

493 log_file.write("O_contigs:\n")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

494 for x in O_nodes_roles:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

495 if "O-1,3,19_not_in_3,10" not in x[0]:#O-1,3,19_not_in_3,10 is just a small size marker

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

496 print x[0].split("___")[-1],x[0].split("__")[0],"blast score:",x[1],"identity%:",str(round(x[2]*100,2))+"%"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

497 log_file.write(x[0].split("___")[-1]+" "+x[0].split("__")[0]+" "+"blast score: "+str(x[1])+"identity%:"+str(round(x[2]*100,2))+"%"+"\n")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

498 print "H_contigs:"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

499 log_file.write("H_contigs:\n")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

500 H_contig_stat=[]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

501 for i in range(len(H_contig_roles)):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

502 x=H_contig_roles[i]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

503 a=0

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

504 for y in Final_list_passed:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

505 if x[1] in y[0] and y[0].startswith(x[0]):

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

506 if "first" in y[0] or "last" in y[0]: #this is the final filter to decide it's fliC or fljB, if can't pass, then can't decide

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

507 for y in Final_list_passed: #it's impossible to has the "first" and "last" allele as prediction, so re-do it

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

508 if x[1] in y[0]:#it's very possible to be third phase allele, so no need to make it must be fliC or fljB

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

509 print x[1],"can't_decide_fliC_or_fljB",y[0].split("_")[1],"blast_score:",y[1],"identity%:",str(round(y[2]*100,2))+"%"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

510 log_file.write(x[1]+" "+x[0]+" "+y[0].split("_")[1]+" "+"blast_score: "+str(y[1])+" identity%:"+str(round(y[2]*100,2))+"%"+"\n")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

511 H_contig_roles[i]="can't decide fliC or fljB, may be third phase"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

512 break

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

513 else:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

514 print x[1],x[0],y[0].split("_")[1],"blast_score:",y[1],"identity%:",str(round(y[2]*100,2))+"%"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

515 log_file.write(x[1]+" "+x[0]+" "+y[0].split("_")[1]+" "+"blast_score: "+str(y[1])+" identity%:"+str(round(y[2]*100,2))+"%"+"\n")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

516 break

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

517 for x in H_contig_roles:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

518 #if multiple choices, temporately select the one with longest length for now, will revise in further change

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

519 if "fliC" == x[0] and int(x[1].split("_")[3])>=fliC_length and x[1] not in O_nodes:#remember to avoid the effect of O-type contig, so should not in O_node list

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

520 fliC_contig=x[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

521 fliC_length=int(x[1].split("_")[3])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

522 elif "fljB" == x[0] and int(x[1].split("_")[3])>=fljB_length and x[1] not in O_nodes:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

523 fljB_contig=x[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

524 fljB_length=int(x[1].split("_")[3])

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

525 for x in Final_list_passed:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

526 if fliC_choice=="-" and "fliC_" in x[0] and fliC_contig in x[0] :

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

527 fliC_choice=x[0].split("_")[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

528 elif fljB_choice=="-" and "fljB_" in x[0] and fljB_contig in x[0]:

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

529 fljB_choice=x[0].split("_")[1]

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

530 elif fliC_choice!="-" and fljB_choice!="-":

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

531 break

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

532 print "\n"

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

533 print "SeqSero Input files:",for_fq,rev_fq

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

534 print "Most possible O antigen:",O_choice

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

535 print "Most possible H1 antigen:",fliC_choice

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

536 print "Most possible H2 antigen:",fljB_choice

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

537 #print Final_list

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

538 ###output

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

539 predict_form,predict_sero,star,star_line,claim=seqsero_from_formula_to_serotypes(O_choice,fliC_choice,fljB_choice,special_gene_list)

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

540 new_file=open("Seqsero_result.txt","w")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

541 new_file.write("Input files:\t"+for_fq+" "+rev_fq+"\n"+"O antigen prediction:\t"+"O-"+O_choice+"\n"+"H1 antigen prediction(fliC):\t"+fliC_choice+"\n"+"H2 antigen prediction(fljB):\t"+fljB_choice+"\n"+"Predicted antigenic profile:\t"+predict_form+"\n"+"Predicted serotype(s):\t"+predict_sero+star+"\n"+star+star_line+claim+"\n")#+##

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

542 new_file.close()

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

543 os.system("cat Seqsero_result.txt")

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

544

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

545 if __name__ == '__main__':

6275272ebcbc planemo upload commit 9b152b4a900a8cd70df992da881c7e3fa00d4e4c-dirty

cstrittmatter

parents:

diff changeset

546 main()

Mercurial > repos > cstrittmatter > seqsero_v2

annotate libs/mapping_and_assembly_hybrid.py @ 0:6275272ebcbc draft