mirror of
https://github.com/galaxyproject/galaxy.git
synced 2026-10-29 17:42:12 +08:00
BLAT wrapper for short read analysis.
include functional test data.
This commit is contained in:
Wen-Yu Chung
@@ -0,0 +1,83 @@
|
||||
#! /usr/bin/python
|
||||
|
||||
import os, sys, tempfile
|
||||
|
||||
nib_file = "/depot/data2/galaxy/alignseq.loc"
|
||||
twobit_file = "/depot/data2/galaxy/twobit.loc"
|
||||
|
||||
def check_nib_file( dbkey ):
|
||||
nib_path = ''
|
||||
nibs = {}
|
||||
for line in open( nib_file ):
|
||||
line = line.rstrip( '\r\n' )
|
||||
if line and not line.startswith( "#" ):
|
||||
fields = line.split()
|
||||
if len( fields ) < 3:
|
||||
continue
|
||||
if ( fields[0] == 'seq' ):
|
||||
nibs[( fields[1] )] = fields[2]
|
||||
if nibs.has_key( dbkey ):
|
||||
nib_path = nibs[( dbkey )]
|
||||
return nib_path
|
||||
|
||||
def check_twobit_file( dbkey ):
|
||||
twobit_path = ''
|
||||
twobits = {}
|
||||
for line in open( twobit_file ):
|
||||
line = line.rstrip( '\r\n' )
|
||||
if line and not line.startswith( "#" ):
|
||||
fields = line.split()
|
||||
if len( fields ) < 2:
|
||||
continue
|
||||
twobits[( fields[0] )] = fields[1]
|
||||
if twobits.has_key( dbkey ):
|
||||
twobit_path = twobits[( dbkey )]
|
||||
return twobit_path
|
||||
|
||||
def __main__():
|
||||
# I/O
|
||||
source_format = sys.argv[1] # 0: dbkey; 1: upload file
|
||||
target_file = sys.argv[2]
|
||||
query_file = sys.argv[3]
|
||||
output_file = sys.argv[4]
|
||||
min_iden = sys.argv[5]
|
||||
tile_size = sys.argv[6]
|
||||
one_off = sys.argv[7]
|
||||
|
||||
all_files = []
|
||||
if (source_format == '0'):
|
||||
# check target genome
|
||||
dbkey = target_file
|
||||
if dbkey == '?':
|
||||
print >> sys.stdout, "No genome build specified. please check your dataset."
|
||||
sys.exit()
|
||||
nib_path = check_nib_file( dbkey )
|
||||
twobit_path = check_twobit_file( dbkey )
|
||||
if not os.path.exists( nib_path ) and not os.path.exists( twobit_path ):
|
||||
print >> sys.stdout, "No sequences are available for %s, request them by reporting this error." % dbkey
|
||||
sys.exit()
|
||||
|
||||
# check the query file, see whether all of them are legitimate sequence
|
||||
if (nib_path):
|
||||
compress_files = os.listdir(nib_path)
|
||||
target_path = nib_path
|
||||
elif (twobit_path):
|
||||
compress_files = twobit_path
|
||||
target_path = ""
|
||||
else:
|
||||
print >> sys.stdout, "Requested genome build has no available sequence."
|
||||
sys.exit()
|
||||
for file in compress_files:
|
||||
file = target_path + '/' + file
|
||||
file = os.path.normpath(file)
|
||||
all_files.append(file)
|
||||
else:
|
||||
all_files = [target_file]
|
||||
|
||||
for detail_file_path in all_files:
|
||||
output_tempfile = tempfile.NamedTemporaryFile().name
|
||||
command = 'blat ' + detail_file_path + ' ' + query_file + ' ' + output_tempfile + ' -oneOff=' + one_off + ' -tileSize=' + tile_size + ' -minIdentity=' + min_iden + ' -mask=lower -noHead -out=pslx 2>&1'
|
||||
os.system(command)
|
||||
os.system('cat %s >> %s' %(output_tempfile, output_file))
|
||||
|
||||
if __name__ == '__main__': __main__()
|
||||
@@ -0,0 +1,69 @@
|
||||
<tool id="blat_wrapper" name="Run Sequence Alignment" version="1.0.0">
|
||||
<description>on short reads against genome builds</description>
|
||||
<command interpreter="python">
|
||||
#if $source.source_select=="database":#blat_wrapper.py 0 $source.dbkey $input_query $output1 $iden $tile_size $one_off
|
||||
#else:#blat_wrapper.py 1 $source.input_target $input_query $output1 $iden $tile_size $one_off
|
||||
#end if
|
||||
</command>
|
||||
<inputs>
|
||||
<conditional name="source">
|
||||
<param name="source_select" type="select" label="Target Source">
|
||||
<option value="database">Genome Build</option>
|
||||
<option value="input_ref">Your Upload File</option>
|
||||
</param>
|
||||
<when value="database">
|
||||
<param name="dbkey" type="genomebuild" label="Genome" />
|
||||
</when>
|
||||
<when value="input_ref">
|
||||
<param name="input_target" type="data" format="fasta" label="Reference Sequence" />
|
||||
</when>
|
||||
</conditional>
|
||||
<param name="input_query" type="data" format="fasta" label="Query Sequence"/>
|
||||
<param name="iden" type="float" size="15" value="90.0" label="Minimal Identity (-minIdentity)" />
|
||||
<param name="tile_size" type="integer" size="15" value="11" label="Minimal Size of Match (-tileSize)" help="for shorter reads, use size = 8"/>
|
||||
<param name="one_off" type="integer" size="15" value="0" label="Number of Mismatch in the Word (-oneOff)" />
|
||||
</inputs>
|
||||
<outputs>
|
||||
<data name="output1" format="tabular"/>
|
||||
</outputs>
|
||||
<tests>
|
||||
<test>
|
||||
<param name="source_select" value="database" />
|
||||
<param name="dbkey" value="eschColi_K12_1" />
|
||||
<param name="input_query" value="blat_wrapper_test1.fa" ftype="fasta"/>
|
||||
<param name="iden" value="90.0" />
|
||||
<param name="tile_size" value="11" />
|
||||
<param name="one_off" value="0" />
|
||||
<output name="output1" file="blat_wrapper_test1.out" />
|
||||
</test>
|
||||
</tests>
|
||||
<help>
|
||||
|
||||
**What it does**
|
||||
|
||||
This tool runs sequence alignment program on your short reads dataset against a genome build.
|
||||
|
||||
-----
|
||||
|
||||
**Example**
|
||||
|
||||
Input as multiple fasta file::
|
||||
|
||||
>seq1
|
||||
TGGTAATGGTGGTTTTTTTTTTTTTTTTTTATTTTT
|
||||
|
||||
Search against ce2 (C. elegans March 2004)
|
||||
|
||||
(This is output type = pslx)::
|
||||
|
||||
25 1 0 0 0 0 0 0 + seq1 36 10 36 chrI 15080483 9704438 9704464 1 26, 10, 9704438, ggttttttttttttttttttattttt, ggtttttttttttttttttttttttt,
|
||||
27 0 0 0 0 0 1 32 + seq1 36 9 36 chrI 15080483 1302536 1302595 2 21,6, 9,30, 1302536,1302589, tggtttttttttttttttttt,attttt, tggtttttttttttttttttt,attttt,
|
||||
|
||||
-----
|
||||
|
||||
.. class:: infomark
|
||||
|
||||
BLAT: Kent, W James, BLAT--the BLAST-like alignment tool. (2002) Genome Research:12(4) 656-664.
|
||||
|
||||
</help>
|
||||
</tool>
|
||||
Reference in New Issue
Block a user