Mercurial > repos > davidvanzessen > shm_csr

annotate baseline/script_xlsx.py @ 98:d714f5ea83d7 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 1a01065a084a817382872154f779b94090a35ebf
author rhpvorderman
date Wed, 10 Jan 2024 12:32:47 +0000
parents 729738462297
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
83
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
1 import xlrd
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
2 import argparse
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
3
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
4 parser = argparse.ArgumentParser()
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
5 parser.add_argument("--input", help="Excel input file containing one or more sheets where column G has the gene annotation, H has the sequence id and J has the sequence")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
6 parser.add_argument("--ref", help="Reference file")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
7 parser.add_argument("--output", help="Output file")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
8
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
9 args = parser.parse_args()
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
10
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
11 gene_column = 6
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
12 id_column = 7
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
13 seq_column = 8
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
14 LETTERS = [x for x in "ABCDEFGHIJKLMNOPQRSTUVWXYZ"]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
15
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
16
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
17 refdic = dict()
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
18 with open(args.ref, 'r') as ref:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
19 currentSeq = ""
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
20 currentId = ""
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
21 for line in ref.readlines():
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
22 if line[0] is ">":
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
23 if currentSeq is not "" and currentId is not "":
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
24 refdic[currentId[1:]] = currentSeq
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
25 currentId = line.rstrip()
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
26 currentSeq = ""
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
27 else:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
28 currentSeq += line.rstrip()
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
29 refdic[currentId[1:]] = currentSeq
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
30
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
31 currentSeq = ""
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
32 currentId = ""
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
33 with xlrd.open_workbook(args.input, 'r') as wb:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
34 with open(args.output, 'a') as o:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
35 for sheet in wb.sheets():
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
36 if sheet.cell(1,gene_column).value.find("IGHV") < 0:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
37 print("Genes not in column " + LETTERS[gene_column] + ", skipping sheet " + sheet.name)
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
38 continue
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
39 o.write(">>>" + sheet.name + "\n")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
40 outputdic = dict()
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
41 for rowindex in range(1, sheet.nrows):
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
42 ref = sheet.cell(rowindex, gene_column).value.replace(">", "")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
43 if ref in outputdic:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
44 outputdic[ref] += [(sheet.cell(rowindex, id_column).value.replace(">", ""), sheet.cell(rowindex, seq_column).value)]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
45 else:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
46 outputdic[ref] = [(sheet.cell(rowindex, id_column).value.replace(">", ""), sheet.cell(rowindex, seq_column).value)]
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
47 #print outputdic
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
48
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
49 for k in list(outputdic.keys()):
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
50 if k in refdic:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
51 o.write(">>" + k + "\n")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
52 o.write(refdic[k] + "\n")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
53 for seq in outputdic[k]:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
54 #print seq
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
55 o.write(">" + seq[0] + "\n")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
56 o.write(seq[1] + "\n")
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
57 else:
729738462297 "planemo upload commit c0ffc68aec5836d5b20b543106493056a87edf57"
rhpvorderman
parents: 0
diff changeset
58 print(k + " not in reference, skipping " + k)