Mercurial > repos > iuc > fasta_stats

annotate fasta-stats.py @ 4:0dbb995c7d35 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
author iuc
date Thu, 18 Nov 2021 20:56:57 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
4
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
1 #!/usr/bin/env python
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
2
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
3
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
4 # python version of fasta-stats with some extra features
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
5 # written by anmol.kiran@gmail.com
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
6 # git: @codemeleon
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
7 # date: 10/11/2021
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
8
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
9 import argparse
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
10 import re
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
11 from os import path
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
12
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
13 import numpy as np
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
14 from Bio import SeqIO
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
15
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
16
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
17 def calculate_NG50(estimated_genome, total_length, sequence_lengths):
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
18 temp = 0
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
19 teoretical_NG50 = estimated_genome / 2.0
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
20 NG50 = 0
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
21 for seq in sequence_lengths:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
22 temp += seq
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
23 if teoretical_NG50 < temp:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
24 NG50 = seq
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
25 break
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
26 return NG50
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
27
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
28
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
29 def run(fasta, stats_output, gaps_output, genome_size):
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
30 """Generates scaffold statistics."""
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
31 if not fasta:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
32 exit("Input file not given.")
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
33 if not path.isfile(fasta):
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
34 exit(f"{fasta} path does not exist.")
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
35
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
36 seq_len = {}
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
37 bases_global = {"A": 0, "N": 0, "T": 0, "C": 0, "G": 0}
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
38 bases_seq = {}
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
39 seq_id_Ngaprange = {}
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
40 nstart = 0
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
41 contigs_len = []
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
42 gap_count = 0
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
43 for seq_record in SeqIO.parse(fasta, "fasta"):
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
44 seq = str(seq_record.seq).upper()
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
45 # print(len(seq))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
46 seq_len[seq_record.id] = len(seq)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
47
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
48 # NOTE: Nucleotide count
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
49 bases_seq[seq_record.id] = {
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
50 "A": seq.count("A"),
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
51 "N": seq.count("N"),
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
52 "T": seq.count("T"),
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
53 "C": seq.count("C"),
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
54 "G": seq.count("G"),
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
55 }
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
56 bases_global["A"] += bases_seq[seq_record.id]["A"]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
57 bases_global["N"] += bases_seq[seq_record.id]["N"]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
58 bases_global["T"] += bases_seq[seq_record.id]["T"]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
59 bases_global["C"] += bases_seq[seq_record.id]["C"]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
60 bases_global["G"] += bases_seq[seq_record.id]["G"]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
61
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
62 # NOTE: Gap count and their range
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
63 range_gen = re.finditer("N+", seq)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
64 n_range = [match.span() for match in range_gen]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
65 for n_rng in n_range:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
66 if n_rng[0] == 0 or n_rng[1] == seq_len[seq_record.id]:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
67 continue
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
68 else:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
69 gap_count += 1
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
70
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
71 # NOTE: Contigs, their lenths from scaffold and their N gap range
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
72 seq_id_Ngaprange[seq_record.id] = n_range
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
73 n_range_len = len(n_range)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
74 if n_range_len > 0:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
75 n_range = (
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
76 [(0, 0)] + n_range + [(seq_len[seq_record.id], seq_len[seq_record.id])]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
77 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
78 for idx in range(n_range_len + 1):
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
79 nstart = n_range[idx][1]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
80 nend = n_range[idx + 1][0]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
81 con_len = nend - nstart
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
82 if con_len:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
83 contigs_len.append(con_len)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
84 else:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
85 contigs_len.append(len(seq))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
86
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
87 # NOTE: Scaffold statistics
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
88 SEQ_LEN_LIST = sorted(seq_len.values(), reverse=True)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
89 scaffold_lens = np.array(SEQ_LEN_LIST)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
90 scaffold_lens_sum = np.cumsum(scaffold_lens)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
91 N50_len = scaffold_lens_sum[-1] * 0.5
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
92 N50_idx = np.where(scaffold_lens_sum > N50_len)[0][0]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
93 N90_len = scaffold_lens_sum[-1] * 0.9
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
94 N90_idx = np.where(scaffold_lens_sum > N90_len)[0][0]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
95 NG50 = calculate_NG50(genome_size, scaffold_lens_sum[-1], scaffold_lens)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
96
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
97 # NOTE: Contig statistics
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
98 seq_len_list = sorted(contigs_len, reverse=True)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
99 contigs_len = np.array(seq_len_list)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
100 contigs_len_sum = np.cumsum(contigs_len)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
101 n50_len = contigs_len_sum[-1] * 0.5
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
102 n50_idx = np.where(contigs_len_sum > n50_len)[0][0]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
103 n90_len = contigs_len_sum[-1] * 0.9
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
104 n90_idx = np.where(contigs_len_sum > n90_len)[0][0]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
105 ng50 = calculate_NG50(genome_size, contigs_len_sum[-1], contigs_len)
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
106
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
107 with open(stats_output, "w") as soutput:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
108 soutput.write("{}\t{}\n".format("Scaffold L50", N50_idx + 1))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
109 soutput.write("{}\t{}\n".format("Scaffold N50", SEQ_LEN_LIST[N50_idx]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
110 soutput.write("{}\t{}\n".format("Scaffold L90", N90_idx + 1))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
111 soutput.write("{}\t{}\n".format("Scaffold N90", SEQ_LEN_LIST[N90_idx]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
112 if genome_size != 0:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
113 soutput.write("{}\t{}\n".format("Scaffold NG50", NG50))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
114 soutput.write("{}\t{}\n".format("Scaffold len_max", SEQ_LEN_LIST[0]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
115 soutput.write("{}\t{}\n".format("Scaffold len_min", SEQ_LEN_LIST[-1]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
116 soutput.write(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
117 "{}\t{}\n".format("Scaffold len_mean", int(np.mean(SEQ_LEN_LIST)))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
118 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
119 soutput.write(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
120 "{}\t{}\n".format("Scaffold len_median", int(np.median(SEQ_LEN_LIST)))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
121 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
122 soutput.write("{}\t{}\n".format("Scaffold len_std", int(np.std(SEQ_LEN_LIST))))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
123 soutput.write("{}\t{}\n".format("Scaffold num_A", bases_global["A"]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
124 soutput.write("{}\t{}\n".format("Scaffold num_T", bases_global["T"]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
125 soutput.write("{}\t{}\n".format("Scaffold num_C", bases_global["C"]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
126 soutput.write("{}\t{}\n".format("Scaffold num_G", bases_global["G"]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
127 soutput.write("{}\t{}\n".format("Scaffold num_N", bases_global["N"]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
128 soutput.write("{}\t{}\n".format("Scaffold num_bp", scaffold_lens_sum[-1]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
129 soutput.write(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
130 "{}\t{}\n".format(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
131 "Scaffold num_bp_not_N", scaffold_lens_sum[-1] - bases_global["N"]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
132 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
133 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
134 soutput.write("{}\t{}\n".format("Scaffold num_seq", len(SEQ_LEN_LIST)))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
135 soutput.write(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
136 "{}\t{:.2f}\n".format(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
137 "Scaffold GC content overall",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
138 (
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
139 (bases_global["G"] + bases_global["C"])
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
140 * 100.0
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
141 / scaffold_lens_sum[-1]
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
142 ),
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
143 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
144 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
145
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
146 soutput.write("{}\t{}\n".format("Contig L50", n50_idx + 1))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
147 soutput.write("{}\t{}\n".format("Contig N50", seq_len_list[n50_idx]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
148 soutput.write("{}\t{}\n".format("Contig L90", n90_idx + 1))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
149 soutput.write("{}\t{}\n".format("Contig N90", seq_len_list[n90_idx]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
150 if genome_size != 0:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
151 soutput.write("{}\t{}\n".format("Contig NG50", ng50))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
152 soutput.write("{}\t{}\n".format("Contig len_max", seq_len_list[0]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
153 soutput.write("{}\t{}\n".format("Contig len_min", seq_len_list[-1]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
154 soutput.write("{}\t{}\n".format("Contig len_mean", int(np.mean(seq_len_list))))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
155 soutput.write(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
156 "{}\t{}\n".format("Contig len_median", int(np.median(seq_len_list)))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
157 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
158 soutput.write("{}\t{}\n".format("Contig len_std", int(np.std(seq_len_list))))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
159 soutput.write("{}\t{}\n".format("Contig num_bp", contigs_len_sum[-1]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
160 soutput.write("{}\t{}\n".format("Contig num_seq", len(contigs_len_sum)))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
161 soutput.write("{}\t{}\n".format("Number of gaps", gap_count))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
162 if gaps_output is not None:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
163 # NOTE: generate gaps statistics file
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
164 with open(gaps_output, "w") as goutput:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
165 for key in seq_id_Ngaprange:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
166 for rng in seq_id_Ngaprange[key]:
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
167 goutput.write("{}\t{}\t{}\n".format(key, rng[0], rng[1]))
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
168
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
169
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
170 if __name__ == "__main__":
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
171 parser = argparse.ArgumentParser()
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
172 parser.add_argument("-f", "--fasta", required=True, help="FASTA file")
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
173 parser.add_argument(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
174 "-z",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
175 "--genome_size",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
176 required=False,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
177 type=int,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
178 help="If provided, the NG50 statistic will be computed",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
179 default=0,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
180 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
181 parser.add_argument(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
182 "-s",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
183 "--stats_output",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
184 required=True,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
185 help="File to store the general statistics",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
186 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
187 parser.add_argument(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
188 "-r",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
189 "--gaps_output",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
190 required=False,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
191 help="File to store the gaps statistics",
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
192 default=None,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
193 )
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
194 args = parser.parse_args()
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
195
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
196 run(
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
197 args.fasta,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
198 args.stats_output,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
199 args.gaps_output,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
200 args.genome_size,
0dbb995c7d35 "planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/fasta_stats/ commit 50f5cce5a8c11001e2c59600a2b99a4243b6d06f"
iuc
parents:
diff changeset
201 )