Mercurial > repos > fubar > egapx_runner

annotate ui/egapx.py @ 1:c8e1543546f8 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4-dirty
author fubar
date Sat, 03 Aug 2024 12:10:13 +0000
parents d9c5c5b87fec
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
1 #!/usr/bin/env python
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
2 # requires pip install pyyaml
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
3 #
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
4 import shlex
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
5 import shutil
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
6 import sys
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
7 import os
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
8 import argparse
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
9 import subprocess
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
10 import tempfile
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
11 import re
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
12 import time
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
13 import datetime
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
14 from collections import defaultdict
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
15 from ftplib import FTP
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
16 from ftplib import FTP_TLS
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
17 import ftplib
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
18 from pathlib import Path
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
19 from typing import List
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
20 from urllib.request import urlopen
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
21 import json
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
22 import sqlite3
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
23 import stat
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
24
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
25 import yaml
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
26
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
27 VERBOSITY_DEFAULT=0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
28 VERBOSITY_QUIET=-1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
29 VERBOSITY_VERBOSE=1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
30
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
31 FTP_EGAP_PROTOCOL = "https"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
32 FTP_EGAP_SERVER = "ftp.ncbi.nlm.nih.gov"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
33 FTP_EGAP_ROOT_PATH = "genomes/TOOLS/EGAP/support_data"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
34 FTP_EGAP_ROOT = f"{FTP_EGAP_PROTOCOL}://{FTP_EGAP_SERVER}/{FTP_EGAP_ROOT_PATH}"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
35 DATA_VERSION = "current"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
36 dataset_taxonomy_url = "https://api.ncbi.nlm.nih.gov/datasets/v2alpha/taxonomy/taxon/"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
37
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
38 user_cache_dir = ''
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
39
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
40 def parse_args(argv):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
41 parser = argparse.ArgumentParser(description="Main script for EGAPx")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
42 group = parser.add_argument_group('run')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
43 group.add_argument("filename", nargs='?', help="YAML file with input: section with at least genome: and reads: parameters")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
44 group.add_argument("-o", "--output", help="Output path", default="")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
45 parser.add_argument("-e", "--executor", help="Nextflow executor, one of docker, singularity, aws, or local (for NCBI internal use only). Uses corresponding Nextflow config file", default="local")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
46 parser.add_argument("-c", "--config-dir", help="Directory for executor config files, default is ./egapx_config. Can be also set as env EGAPX_CONFIG_DIR", default="")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
47 parser.add_argument("-w", "--workdir", help="Working directory for cloud executor", default="")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
48 parser.add_argument("-r", "--report", help="Report file prefix for report (.report.html) and timeline (.timeline.html) files, default is in output directory", default="")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
49 parser.add_argument("-n", "--dry-run", action="store_true", default=False)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
50 parser.add_argument("-st", "--stub-run", action="store_true", default=False)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
51 parser.add_argument("-so", "--summary-only", help="Print result statistics only if available, do not compute result", action="store_true", default=False)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
52 group = parser.add_argument_group('download')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
53 group.add_argument("-dl", "--download-only", help="Download external files to local storage, so that future runs can be isolated", action="store_true", default=False)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
54 parser.add_argument("-lc", "--local-cache", help="Where to store the downloaded files", default="")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
55 parser.add_argument("-q", "--quiet", dest='verbosity', action='store_const', const=VERBOSITY_QUIET, default=VERBOSITY_DEFAULT)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
56 parser.add_argument("-v", "--verbose", dest='verbosity', action='store_const', const=VERBOSITY_VERBOSE, default=VERBOSITY_DEFAULT)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
57 parser.add_argument("-fn", "--func_name", help="func_name", default="")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
58 return parser.parse_args(argv[1:])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
59
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
60
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
61 class FtpDownloader:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
62 def __init__(self):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
63 self.ftp = None
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
64
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
65 def connect(self, host):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
66 self.host = host
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
67 self.reconnect()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
68
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
69 def reconnect(self):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
70 self.ftp = FTP(self.host)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
71 self.ftp.login()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
72 self.ftp.set_debuglevel(0)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
73
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
74 ##ftp_types = set()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
75 def download_ftp_file(self, ftp_name, local_path):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
76 # print(f"file: {ftp_name}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
77 # print(f"f: { os.path.dirname(local_path)}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
78
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
79 os.makedirs(os.path.dirname(local_path), exist_ok=True)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
80 try:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
81 with open(local_path, 'wb') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
82 self.ftp.retrbinary("RETR {0}".format(ftp_name), f.write)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
83 # print("downloaded: {0}".format(local_path))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
84 return True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
85 except FileNotFoundError:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
86 print("FAILED FNF: {0}".format(local_path))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
87 except EOFError:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
88 print("FAILED EOF: {0}".format(local_path))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
89 time.sleep(1)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
90 self.reconnect()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
91 print("retrying...")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
92 r = self.download_ftp_file(ftp_name, local_path)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
93 print("downloaded on retry: {0}".format(local_path))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
94 return r
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
95 except BrokenPipeError:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
96 print("FAILED BP: {0}".format(local_path))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
97 except IsADirectoryError:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
98 ## same as 550 but pre-exists
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
99 ## os.remove(local_path)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
100 return 550
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
101 except ftplib.error_perm:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
102 ## ftplib.error_perm: 550 genomes/TOOLS/EGAP/ortholog_references/9606/current: Not a regular file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
103 ## its a symlink to a dir.
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
104 os.remove(local_path)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
105 return 550
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
106 return False
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
107
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
108 # item: ('Eublepharis_macularius', {'modify': '20240125225739', 'perm': 'fle', 'size': '4096', 'type': 'dir', 'unique': '6CU599079F', 'unix.group': '562', 'unix.mode': '0444', 'unix.owner': '14'}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
109 def should_download_file(self, ftp_item, local_name):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
110 metadata = ftp_item[1]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
111 ftp_modify = datetime.datetime.strptime(metadata['modify'], '%Y%m%d%H%M%S')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
112 ftp_size = int(metadata['size'])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
113 ftp_type = metadata['type']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
114
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
115 local_stat = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
116 try:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
117 local_stat = os.stat(local_name)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
118 except FileNotFoundError:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
119 return True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
120 except NotADirectoryError:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
121 return True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
122 local_is_dir = stat.S_ISDIR(local_stat.st_mode)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
123 #print(f"item: {ftp_item}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
124 #print(f"stat: {local_stat}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
125
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
126 local_stat_dt = datetime.datetime.fromtimestamp(local_stat.st_mtime)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
127
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
128 #print(f"should_dl: {ftp_size != local_stat.st_size} {ftp_modify > local_stat_dt} ")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
129
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
130 if (ftp_type == 'file' and ftp_size != local_stat.st_size) or (ftp_type=='OS.unix=symlink' and ftp_size >= local_stat.st_size):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
131 return True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
132
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
133 if ftp_modify > local_stat_dt:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
134 return True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
135
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
136 return False
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
137
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
138 ## types
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
139 ## cdir and pdir: . and ..: do nothing
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
140 ## file: a file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
141 ## dir: a dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
142 ## OS.unix=symlink: symlink to a file, treat like a file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
143 def download_ftp_dir(self, ftp_path, local_path):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
144 """ replicates a directory on an ftp server recursively """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
145 for ftp_item in self.ftp.mlsd(ftp_path):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
146 # print(f"ftp_item: {ftp_item}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
147 ##print(f" name: {ftp_item[0]}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
148 ##print(f" type: {ftp_item[1]['type']}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
149 name = ftp_item[0]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
150 next_ftp_name="/".join([ftp_path,name])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
151 next_local_name=os.sep.join([local_path,name])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
152 ftp_type = ftp_item[1]['type']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
153 ##ftp_types.add(ftp_type)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
154 if ftp_type=='dir':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
155 self.download_ftp_dir(next_ftp_name, next_local_name)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
156 elif ftp_type=='file' or ftp_type=='OS.unix=symlink':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
157 if self.should_download_file(ftp_item, next_local_name):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
158 r = self.download_ftp_file(next_ftp_name, next_local_name)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
159 if r == 550:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
160 self.download_ftp_dir(next_ftp_name, next_local_name)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
161 ##time.sleep(0.1)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
162 ## else: . or .. do nothing
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
163
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
164
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
165 def download_egapx_ftp_data(local_cache_dir):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
166 global user_cache_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
167 manifest_url = f"{FTP_EGAP_ROOT}/{DATA_VERSION}.mft"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
168 manifest = urlopen(manifest_url)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
169 manifest_path = f"{user_cache_dir}/{DATA_VERSION}.mft"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
170 manifest_list = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
171 for line in manifest:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
172 line = line.decode("utf-8").strip()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
173 if not line or line[0] == '#':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
174 continue
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
175 manifest_list.append(line)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
176 print(f"Downloading {line}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
177 ftpd = FtpDownloader()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
178 ftpd.connect(FTP_EGAP_SERVER)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
179 ftpd.download_ftp_dir(FTP_EGAP_ROOT_PATH+f"/{line}", f"{local_cache_dir}/{line}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
180 if user_cache_dir:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
181 with open(manifest_path, 'wt') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
182 for line in manifest_list:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
183 f.write(f"{line}\n")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
184 return 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
185
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
186
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
187 def repackage_inputs(run_inputs):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
188 "Repackage input parameters into 'input' key if not already there"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
189 if 'input' in run_inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
190 return run_inputs
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
191 new_inputs = {}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
192 new_inputs['input'] = {}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
193 for key in run_inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
194 if key not in { 'tasks', 'output' }:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
195 new_inputs['input'][key] = run_inputs[key]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
196 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
197 new_inputs[key] = run_inputs[key]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
198 return new_inputs
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
199
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
200
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
201 def convert_value(value):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
202 "Convert paths to absolute paths when possible, look into objects and lists"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
203 if isinstance(value, dict):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
204 return {k: convert_value(v) for k, v in value.items()}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
205 elif isinstance(value, list):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
206 return [convert_value(v) for v in value]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
207 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
208 if value == '' or re.match(r'[a-z0-9]{2,5}://', value) or not os.path.exists(value):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
209 # do not convert - this is a URL or empty string or non-file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
210 return value
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
211 # convert to absolute path
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
212 return str(Path(value).absolute())
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
213
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
214
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
215 path_inputs = { 'genome', 'hmm', 'softmask', 'reads_metadata', 'organelles', 'proteins', 'reads', 'rnaseq_alignments', 'protein_alignments' }
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
216 def convert_paths(run_inputs):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
217 "Convert paths to absolute paths where appropriate"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
218 input_root = run_inputs['input']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
219 for key in input_root:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
220 if key in path_inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
221 if isinstance(input_root[key], list):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
222 input_root[key] = [convert_value(v) for v in input_root[key]]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
223 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
224 input_root[key] = convert_value(input_root[key])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
225 if 'output' in run_inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
226 run_inputs['output'] = convert_value(run_inputs['output'])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
227
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
228
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
229 def prepare_reads(run_inputs):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
230 """Reformat reads input to be in 'fromPairs' format expected by egapx, i.e. [sample_id, [read1, read2]]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
231 Generate reads metadata file with minimal information - paired/unpaired and valid for existing libraries"""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
232 if 'reads' not in run_inputs['input'] or 'output' not in run_inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
233 return
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
234 prefixes = defaultdict(list)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
235 reads = run_inputs['input']['reads']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
236 if type(reads) == str:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
237 del run_inputs['input']['reads']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
238 run_inputs['input']['reads_query'] = reads
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
239 return
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
240 # Create fake metadata file for reads containing only one meaningful information piece - pairedness
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
241 # Have an entry in this file only for SRA runs - files with prefixes SRR, ERR, and DRR and digits
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
242 # Non-SRA reads are handled by rnaseq_collapse internally
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
243 has_files = False
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
244 for rf in run_inputs['input']['reads']:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
245 if type(rf) == str:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
246 name = Path(rf).parts[-1]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
247 mo = re.match(r'([^._]+)', name)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
248 if mo and mo.group(1) != name:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
249 has_files = True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
250 prefixes[mo.group(1)].append(rf)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
251 elif type(rf) == list:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
252 has_files = True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
253 names = list(map(lambda x: re.match(r'([^.]+)', Path(x).parts[-1]).group(1), rf))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
254 # Find common prefix
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
255 names.sort()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
256 if len(names) == 1:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
257 sample_id = names[0]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
258 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
259 for i in range(min(len(names[0]), len(names[-1]))):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
260 if names[0][i] != names[-1][i]:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
261 break
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
262 sample_id = names[0][0:i]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
263 # Dont end prefix with . or _
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
264 i = len(sample_id) - 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
265 while i > 0 and sample_id[-1] in "._":
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
266 sample_id = sample_id[:-1]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
267 i -= 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
268 prefixes[sample_id] = rf
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
269 if has_files: # len(prefixes):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
270 # Always create metadata file even if it's empty
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
271 output = run_inputs['output']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
272 with tempfile.NamedTemporaryFile(mode='w', delete=False, dir=output, prefix='egapx_reads_metadata_', suffix='.tsv') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
273 for k, v in prefixes.items():
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
274 if re.fullmatch(r'([DES]RR[0-9]+)', k):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
275 paired = 'paired' if len(v) == 2 else 'unpaired'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
276 # SRR9005248 NA paired 2 2 NA NA NA NA NA NA NA 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
277 rec = "\t".join([k, 'NA', paired, '2', '2', 'NA', 'NA', 'NA', 'NA', 'NA', 'NA', 'NA', '0'])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
278 f.write(rec + '\n')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
279 f.flush()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
280 run_inputs['input']['reads_metadata'] = f.name
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
281 run_inputs['input']['reads'] = [ [k, v] for k, v in prefixes.items() ]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
282 elif reads:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
283 del run_inputs['input']['reads']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
284 run_inputs['input']['reads_query'] = "[Accession] OR ".join(reads) + "[Accession]"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
285
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
286
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
287 def expand_and_validate_params(run_inputs):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
288 """ Expand implicit parameters and validate inputs
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
289 Args:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
290 run_inputs (dict): Input parameters
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
291 Returns:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
292 bool: True if parameters are valid, False otherwise
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
293 """
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
294 inputs = run_inputs['input']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
295
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
296 if 'taxid' not in inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
297 print("ERROR: Missing parameter: 'taxid'")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
298 return False
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
299
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
300 taxid = inputs['taxid']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
301
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
302 if 'genome' not in inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
303 print("ERROR: Missing parameter: 'genome'")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
304 return False
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
305
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
306 # Check for proteins input and if empty or no input at all, add closest protein bag
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
307 if 'proteins' not in inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
308 proteins = get_closest_protein_bag(taxid)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
309 if not proteins:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
310 # Proteins are not specified explicitly and not found by taxid
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
311 print(f"ERROR: Proteins are not found for tax id {inputs['taxid']}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
312 print(" This organism is not supported by current NCBI protein database")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
313 print(" You can specify proteins manually using 'proteins' parameter")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
314 return False
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
315 inputs['proteins'] = proteins
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
316
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
317 has_rnaseq = 'reads' in inputs or 'reads_ids' in inputs or 'reads_query' in inputs
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
318 if not has_rnaseq:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
319 if inputs['proteins']:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
320 print("WARNING: It is strongly advised to specify RNA-seq reads using 'reads' parameter\n")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
321 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
322 print("ERROR: Either proteins or RNA-seq reads should be provided for annotation")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
323 return False
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
324
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
325 if 'hmm' not in inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
326 best_taxid, best_hmm = get_closest_hmm(taxid)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
327 inputs['hmm'] = best_hmm
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
328 inputs['hmm_taxid'] = best_taxid
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
329 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
330 # We assume the user knows what they're doing and the training is not necessary
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
331 inputs['hmm_taxid'] = taxid
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
332
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
333 if 'max_intron' not in inputs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
334 max_intron, genome_size_threshold = get_max_intron(taxid)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
335 inputs['max_intron'] = max_intron
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
336 inputs['genome_size_threshold'] = genome_size_threshold
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
337 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
338 # Given max_intron is a hard limit, no further calculation is necessary
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
339 inputs['genome_size_threshold'] = 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
340
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
341 return True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
342
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
343
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
344 def manage_workdir(args):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
345 workdir_file = f"work_dir_{args.executor}.last"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
346 if args.workdir:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
347 os.environ['NXF_WORK'] = args.workdir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
348 with open(workdir_file, 'w') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
349 f.write(args.workdir)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
350 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
351 if os.path.exists(workdir_file):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
352 with open(workdir_file) as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
353 os.environ['NXF_WORK'] = f.read().strip()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
354 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
355 if args.executor == 'aws':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
356 print("Work directory not set, use -w at least once")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
357 return False
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
358 return True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
359
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
360
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
361 def get_cache_dir():
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
362 global user_cache_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
363 if user_cache_dir and os.path.exists(user_cache_dir):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
364 return user_cache_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
365 ##elif os.path.exists("/am/ftp-genomes/TOOLS/EGAP"):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
366 ## return "/am/ftp-genomes/TOOLS/EGAP"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
367 return ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
368
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
369
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
370 data_version_cache = {}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
371 def get_versioned_path(subsystem, filename):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
372 global data_version_cache
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
373 global user_cache_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
374 if not data_version_cache:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
375 manifest_path = f"{user_cache_dir}/{DATA_VERSION}.mft"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
376 if user_cache_dir and os.path.exists(manifest_path):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
377 with open(manifest_path, 'rt') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
378 for line in f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
379 line = line.strip()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
380 if not line or line[0] == '#':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
381 continue
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
382 parts = line.split('/')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
383 if len(parts) == 2:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
384 data_version_cache[parts[0]] = parts[1]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
385 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
386 manifest_url = f"{FTP_EGAP_ROOT}/{DATA_VERSION}.mft"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
387 manifest = urlopen(manifest_url)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
388 manifest_list = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
389 for line in manifest:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
390 line = line.decode("utf-8").strip()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
391 if not line or line[0] == '#':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
392 continue
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
393 parts = line.split('/')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
394 if len(parts) == 2:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
395 data_version_cache[parts[0]] = parts[1]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
396 manifest_list.append(line)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
397 if user_cache_dir:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
398 with open(manifest_path, 'wt') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
399 for line in manifest_list:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
400 f.write(f"{line}\n")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
401
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
402 if subsystem not in data_version_cache:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
403 return os.path.join(subsystem, filename)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
404 version = data_version_cache[subsystem]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
405 return os.path.join(subsystem, version, filename)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
406
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
407
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
408 # Get file path for subsystem, either from cache or from remote server
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
409 def get_file_path(subsystem, filename):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
410 cache_dir = get_cache_dir()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
411 vfn = get_versioned_path(subsystem, filename)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
412 file_path = os.path.join(cache_dir, vfn)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
413 file_url = f"{FTP_EGAP_ROOT}/{vfn}"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
414 if os.path.exists(file_path):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
415 return file_path
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
416 return file_url
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
417
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
418
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
419 def get_config(script_directory, args):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
420 config_file = ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
421 config_dir = args.config_dir if args.config_dir else os.environ.get("EGAPX_CONFIG_DIR")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
422 if not config_dir:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
423 config_dir = Path(os.getcwd()) / "egapx_config"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
424 if not Path(config_dir).is_dir():
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
425 # Create directory and copy executor config files there
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
426 from_dir = Path(script_directory) / 'assets' / 'config' / 'executor'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
427 if args.verbosity >= VERBOSITY_VERBOSE:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
428 print(f"Copy config files from {from_dir} to {config_dir}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
429 if not args.dry_run:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
430 os.mkdir(config_dir)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
431 ld = os.listdir(from_dir)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
432 for f in ld:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
433 shutil.copy2(from_dir / f, config_dir)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
434 print(f"Edit config files in {config_dir} to reflect your actual configuration, then repeat the command")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
435 return ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
436 config_file = Path(config_dir) / (args.executor + '.config')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
437 if not config_file.is_file():
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
438 print(f"Executor {args.executor} not supported")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
439 return ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
440 default_configs = [ "default.config" ]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
441 with open(str(config_file), 'r') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
442 config_txt = f.read().replace('\n', ' ')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
443 # Check whether the config sets the container
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
444 mo = re.search(r"process.+container *=", config_txt)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
445 if not mo:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
446 default_configs.append("docker_image.config")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
447 # Check whether the config specifies proccess memory or CPUs
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
448 mo = re.search(r"process.+(memory|cpus) *=", config_txt)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
449 if not mo:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
450 default_configs.append("process_resources.config")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
451
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
452 # Add mandatory configs
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
453 config_files = [str(config_file)]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
454 for cf in default_configs:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
455 config_files.append(os.path.join(script_directory, "assets/config", cf))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
456 return ",".join(config_files)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
457
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
458
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
459 lineage_cache = {}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
460 def get_lineage(taxid):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
461 global lineage_cache
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
462 if not taxid:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
463 return []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
464 if taxid in lineage_cache:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
465 return lineage_cache[taxid]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
466 # Try cached taxonomy database
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
467 taxonomy_db_name = os.path.join(get_cache_dir(), get_versioned_path("taxonomy", "taxonomy4blast.sqlite3"))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
468 if os.path.exists(taxonomy_db_name):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
469 conn = sqlite3.connect(taxonomy_db_name)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
470 if conn:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
471 c = conn.cursor()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
472 lineage = [taxid]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
473 cur_taxid = taxid
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
474 while cur_taxid != 1:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
475 c.execute("SELECT parent FROM TaxidInfo WHERE taxid = ?", (cur_taxid,))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
476 cur_taxid = c.fetchone()[0]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
477 lineage.append(cur_taxid)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
478 lineage.reverse()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
479 lineage_cache[taxid] = lineage
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
480 return lineage
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
481
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
482 # Fallback to API
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
483 taxon_json_file = urlopen(dataset_taxonomy_url+str(taxid))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
484 taxon = json.load(taxon_json_file)["taxonomy_nodes"][0]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
485 lineage = taxon["taxonomy"]["lineage"]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
486 lineage.append(taxon["taxonomy"]["tax_id"])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
487 lineage_cache[taxid] = lineage
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
488 return lineage
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
489
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
490
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
491 def get_tax_file(subsystem, tax_path):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
492 vfn = get_versioned_path(subsystem, tax_path)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
493 taxids_path = os.path.join(get_cache_dir(), vfn)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
494 taxids_url = f"{FTP_EGAP_ROOT}/{vfn}"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
495 taxids_file = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
496 if os.path.exists(taxids_path):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
497 with open(taxids_path, "rb") as r:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
498 taxids_file = r.readlines()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
499 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
500 taxids_file = urlopen(taxids_url)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
501 return taxids_file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
502
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
503 def get_closest_protein_bag(taxid):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
504 if not taxid:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
505 return ''
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
506
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
507 taxids_file = get_tax_file("target_proteins", "taxid.list")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
508 taxids_list = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
509 for line in taxids_file:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
510 line = line.decode("utf-8").strip()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
511 if len(line) == 0 or line[0] == '#':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
512 continue
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
513 parts = line.split('\t')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
514 if len(parts) > 0:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
515 t = parts[0]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
516 taxids_list.append(int(t))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
517
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
518 lineage = get_lineage(taxid)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
519
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
520 best_taxid = None
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
521 best_score = 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
522 for t in taxids_list:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
523 try:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
524 pos = lineage.index(t)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
525 except ValueError:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
526 continue
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
527 if pos > best_score:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
528 best_score = pos
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
529 best_taxid = t
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
530
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
531 if best_score == 0:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
532 return ''
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
533 # print(best_taxid, best_score)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
534 return get_file_path("target_proteins", f"{best_taxid}.faa.gz")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
535
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
536
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
537 def get_closest_hmm(taxid):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
538 if not taxid:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
539 return 0, ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
540
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
541 taxids_file = get_tax_file("gnomon", "hmm_parameters/taxid.list")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
542
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
543 taxids_list = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
544 lineages = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
545 for line in taxids_file:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
546 parts = line.decode("utf-8").strip().split('\t')
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
547 if len(parts) > 0:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
548 t = parts[0]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
549 taxids_list.append(t)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
550 if len(parts) > 1:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
551 l = map(lambda x: int(x) if x[-1] != ';' else int(x[:-1]), parts[1].split())
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
552 lineages.append((int(t), list(l)+[int(t)]))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
553
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
554 #if len(lineages) < len(taxids_list):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
555 # taxonomy_json_file = urlopen(dataset_taxonomy_url+','.join(taxids_list))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
556 # taxonomy = json.load(taxonomy_json_file)["taxonomy_nodes"]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
557 # lineages = [ (t["taxonomy"]["tax_id"], t["taxonomy"]["lineage"] + [t["taxonomy"]["tax_id"]]) for t in taxonomy ]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
558
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
559 lineage = get_lineage(taxid)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
560
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
561 best_lineage = None
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
562 best_taxid = None
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
563 best_score = 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
564 for (t, l) in lineages:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
565 pos1 = 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
566 last_match = 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
567 for pos in range(len(lineage)):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
568 tax_id = lineage[pos]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
569 while tax_id != l[pos1]:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
570 if pos1 + 1 < len(l):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
571 pos1 += 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
572 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
573 break
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
574 if tax_id == l[pos1]:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
575 last_match = pos1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
576 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
577 break
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
578 if last_match > best_score:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
579 best_score = last_match
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
580 best_taxid = t
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
581 best_lineage = l
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
582
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
583 if best_score == 0:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
584 return 0, ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
585 # print(best_lineage)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
586 # print(best_taxid, best_score)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
587 return best_taxid, get_file_path("gnomon", f"hmm_parameters/{best_taxid}.params")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
588
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
589
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
590 PLANTS=33090
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
591 VERTEBRATES=7742
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
592 def get_max_intron(taxid):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
593 if not taxid:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
594 return 0, 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
595 lineage = get_lineage(taxid)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
596 if PLANTS in lineage:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
597 return 300000, 3000000000
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
598 elif VERTEBRATES in lineage:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
599 return 1200000, 2000000000
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
600 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
601 return 600000, 500000000
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
602
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
603
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
604 def to_dict(x: List[str]):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
605 d = {}
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
606 s = len(x)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
607 i = 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
608 while i < s:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
609 el = x[i]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
610 i += 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
611 if el and el[0] == '-':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
612 if i < s:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
613 v = x[i]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
614 if v and (v[0] != '-' or (v[0] == '-' and ' ' in v)):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
615 d[el] = v
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
616 i += 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
617 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
618 d[el] = ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
619 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
620 d[el] = ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
621 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
622 d[el] = ""
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
623 return d
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
624
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
625 def merge_params(task_params, run_inputs):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
626 # Walk over the keys in run_inputs and merge them into task_params
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
627 for k in run_inputs.keys():
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
628 if k in task_params:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
629 if isinstance(task_params[k], dict) and isinstance(run_inputs[k], dict):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
630 task_params[k] = merge_params(task_params[k], run_inputs[k])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
631 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
632 task_params_dict = to_dict(shlex.split(task_params[k]))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
633 run_inputs_dict = to_dict(shlex.split(run_inputs[k]))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
634 task_params_dict.update(run_inputs_dict)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
635 task_params_list = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
636 for k1, v in task_params_dict.items():
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
637 task_params_list.append(k1)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
638 if v:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
639 task_params_list.append(v)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
640 task_params[k] = shlex.join(task_params_list)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
641 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
642 task_params[k] = run_inputs[k]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
643 return task_params
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
644
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
645
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
646 def print_statistics(output):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
647 accept_gff = Path(output) / 'accept.gff'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
648 print(f"Statistics for {accept_gff}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
649 counter = defaultdict(int)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
650 if accept_gff.exists():
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
651 with open(accept_gff, 'rt') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
652 for line in f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
653 line = line.strip()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
654 if not line or line[0] == '#':
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
655 continue
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
656 parts = line.split()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
657 if len(parts) < 3:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
658 continue
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
659 counter[parts[2]] += 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
660 keys = list(counter.keys())
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
661 keys.sort()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
662 for k in keys:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
663 print(f"{k:12s} {counter[k]}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
664
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
665
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
666 def main(argv):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
667 "Main script for EGAPx"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
668 #warn user that this is an alpha release
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
669 print("\n!!WARNING!!\nThis is an alpha release with limited features and organism scope to collect initial feedback on execution. Outputs are not yet complete and not intended for production use.\n")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
670
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
671 # Parse command line
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
672 args = parse_args(argv)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
673
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
674 global user_cache_dir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
675 if args.local_cache:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
676 # print(f"Local cache: {args.local_cache}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
677 user_cache_dir = args.local_cache
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
678 if args.download_only:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
679 if args.local_cache:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
680 if not args.dry_run:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
681 # print(f"Download only: {args.download_only}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
682 os.makedirs(args.local_cache, exist_ok=True)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
683 download_egapx_ftp_data(args.local_cache)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
684 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
685 print(f"Download only to {args.local_cache}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
686 return 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
687 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
688 print("Local cache not set")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
689 return 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
690 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
691 # Check that input and output set
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
692 if not args.filename or not args.output:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
693 print("Input file and output directory must be set")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
694 return 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
695
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
696 packaged_distro = bool(getattr(sys, '_MEIPASS', ''))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
697 script_directory = getattr(sys, '_MEIPASS', os.path.abspath(os.path.dirname(__file__)))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
698 config_file = get_config(script_directory, args)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
699 if not config_file:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
700 return 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
701
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
702 # Check for workdir, set if not set, and manage last used workdir
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
703 if not manage_workdir(args):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
704 return 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
705
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
706 files_to_delete = []
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
707
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
708 # Read default task parameters into a dict
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
709 task_params = yaml.safe_load(open(Path(script_directory) / 'assets' / 'default_task_params.yaml', 'r'))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
710 run_inputs = repackage_inputs(yaml.safe_load(open(args.filename, 'r')))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
711
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
712 if not expand_and_validate_params(run_inputs):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
713 return 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
714
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
715 # Command line overrides manifest input
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
716 if args.output:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
717 run_inputs['output'] = args.output
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
718
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
719 # Convert inputs to absolute paths
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
720 convert_paths(run_inputs)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
721
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
722 # Create output directory if needed
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
723 os.makedirs(run_inputs['output'], exist_ok=True)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
724
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
725 if args.summary_only:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
726 print_statistics(run_inputs['output'])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
727 return 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
728
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
729 # Reformat reads into pairs in fromPairs format and add reads_metadata.tsv file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
730 prepare_reads(run_inputs)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
731
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
732
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
733 ##if True or args.download_only:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
734 ## with open("./dumped_input.yaml", 'w') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
735 ## yaml.dump(run_inputs, f)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
736 ## f.flush()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
737 ##return 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
738
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
739 # Add to default task parameters, if input file has some task parameters they will override the default
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
740 task_params = merge_params(task_params, run_inputs)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
741
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
742 # Move output from YAML file to arguments to have more transparent Nextflow log
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
743 output = task_params['output']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
744 del task_params['output']
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
745
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
746 if args.func_name:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
747 task_params['func_name'] = args.func_name
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
748
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
749 # Run nextflow process
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
750 if args.verbosity >= VERBOSITY_VERBOSE:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
751 task_params['verbose'] = True
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
752 print("Nextflow inputs:")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
753 print(yaml.dump(task_params))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
754 if 'reads_metadata' in run_inputs['input']:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
755 print("Reads metadata:")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
756 with open(run_inputs['input']['reads_metadata'], 'r') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
757 print(f.read())
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
758 if packaged_distro:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
759 main_nf = Path(script_directory) / 'nf' / 'ui.nf'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
760 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
761 main_nf = Path(script_directory) / '..' / 'nf' / 'ui.nf'
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
762 nf_cmd = ["nextflow", "-C", config_file, "-log", f"{output}/nextflow.log", "run", main_nf, "--output", output]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
763 if args.stub_run:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
764 nf_cmd += ["-stub-run", "-profile", "stubrun"]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
765 if args.report:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
766 nf_cmd += ["-with-report", f"{args.report}.report.html", "-with-timeline", f"{args.report}.timeline.html"]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
767 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
768 nf_cmd += ["-with-report", f"{output}/run.report.html", "-with-timeline", f"{output}/run.timeline.html"]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
769
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
770 nf_cmd += ["-with-trace", f"{output}/run.trace.txt"]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
771 # if output directory does not exist, it will be created
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
772 if not os.path.exists(output):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
773 os.makedirs(output)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
774 params_file = Path(output) / "run_params.yaml"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
775 nf_cmd += ["-params-file", str(params_file)]
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
776
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
777 if args.dry_run:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
778 print(" ".join(map(str, nf_cmd)))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
779 else:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
780 with open(params_file, 'w') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
781 yaml.dump(task_params, f)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
782 f.flush()
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
783 if args.verbosity >= VERBOSITY_VERBOSE:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
784 print(" ".join(map(str, nf_cmd)))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
785 resume_file = Path(output) / "resume.sh"
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
786 with open(resume_file, 'w') as f:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
787 f.write("#!/bin/bash\n")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
788 f.write(" ".join(map(str, nf_cmd)))
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
789 f.write(" -resume")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
790 if os.environ.get('NXF_WORK'):
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
791 f.write(" -work-dir " + os.environ['NXF_WORK'])
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
792 f.write("\n")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
793 try:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
794 subprocess.run(nf_cmd, check=True, capture_output=(args.verbosity <= VERBOSITY_QUIET), text=True)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
795 except subprocess.CalledProcessError as e:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
796 print(e.stderr)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
797 print(f"To resume execution, run: sh {resume_file}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
798 if files_to_delete:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
799 print(f"Don't forget to delete file(s) {' '.join(files_to_delete)}")
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
800 return 1
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
801 if not args.dry_run and not args.stub_run:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
802 print_statistics(output)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
803 # TODO: Use try-finally to delete the metadata file
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
804 for f in files_to_delete:
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
805 os.unlink(f)
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
806
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
807 return 0
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
808
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
809 if __name__ == "__main__":
d9c5c5b87fec planemo upload for repository https://github.com/ncbi/egapx commit 8173d01b08d9a91c9ec5f6cb50af346edc8020c4
fubar
parents:
diff changeset
810 sys.exit(main(sys.argv))