opensource/galaxy
3
0
Fork 0
mirror of https://github.com/galaxyproject/galaxy.git synced 2026-09-24 16:30:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #14813 from gregvonkuster/psl_datatype

Browse Source 
Add support for the PSL data format
This commit is contained in:
Marius van den Beek
2022-12-08 16:35:06 +01:00
committed by GitHub
parent 3e4411b8f5 771934d18f
commit c121e2cc25
5 changed files with 140 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
lib/galaxy/config/sample/datatypes_conf.xml.sample
+2
View File
@@ -398,6 +398,7 @@
<converter file="interval_to_bgzip_converter.xml" target_datatype="bgzip"/>
<converter file="interval_to_tabix_converter.xml" target_datatype="tabix" depends_on="bgzip"/>
</datatype>
<datatype extension="psl" type="galaxy.datatypes.tabular:Psl" display_in_upload="true"/>
<datatype extension="obo" type="galaxy.datatypes.text:Obo" mimetype="text/html" display_in_upload="true"/>
<datatype extension="owl" type="galaxy.datatypes.xml:Owl" mimetype="text/html" display_in_upload="true"/>
<datatype extension="png" type="galaxy.datatypes.images:Png" mimetype="image/png" display_in_upload="true"/>
@@ -1125,6 +1126,7 @@
<sniffer type="galaxy.datatypes.interval:Gff"/>
<sniffer type="galaxy.datatypes.interval:Gff3"/>
<sniffer type="galaxy.datatypes.tabular:Pileup"/>
<sniffer type="galaxy.datatypes.tabular:Psl"/>
<sniffer type="galaxy.datatypes.text:Paf"/>
<sniffer type="galaxy.datatypes.interval:Interval"/>
<sniffer type="galaxy.datatypes.tabular:Sam"/>
lib/galaxy/datatypes/sniff.py
+6
View File
@@ -298,6 +298,12 @@ def guess_ext(fname_or_file_prefix: Union[str, "FilePrefix"], sniff_order, is_bi
>>> fname = get_test_fname('megablast_xml_parser_test1.blastxml')
>>> guess_ext(fname, sniff_order)
'blastxml'
>>> fname = get_test_fname('1.psl')
>>> guess_ext(fname, sniff_order)
'psl'
>>> fname = get_test_fname('2.psl')
>>> guess_ext(fname, sniff_order)
'psl'
>>> fname = get_test_fname('interval.interval')
>>> guess_ext(fname, sniff_order)
'interval'
lib/galaxy/datatypes/tabular.py
+121
View File
@@ -1798,3 +1798,124 @@ class CMAP(TabularData):
dataset.metadata.column_types = cleaned_column_types
dataset.metadata.columns = number_of_columns
dataset.metadata.delimiter = "\t"
@build_sniff_from_prefix
class Psl(Tabular):
"""Tab delimited data in psl format."""
edam_format = "format_3007"
file_ext = "psl"
line_class = "assemblies"
data_sources = {"data": "tabix"}
def __init__(self, **kwd):
"""Initialize psl datatype"""
super().__init__(**kwd)
self.column_names = [
"matches",
"misMatches",
"repMatches",
"nCount",
"qNumInsert",
"qBaseInsert",
"tNumInsert",
"tBaseInsert",
"strand",
"qName",
"qSize",
"qStart",
"qEnd",
"tName",
"tSize",
"tStart",
"tEnd",
"blockCount",
"blockSizes",
"qStarts",
"tStarts",
]
def sniff_prefix(self, file_prefix: FilePrefix):
"""
PSL lines represent alignments, and are typically generated
by BLAT. Each line consists of 21 required fields, and track
lines may optionally be used to provide more information.
Fields are tab-separated, and all 21 are required.
Although not part of the formal PSL specification, track lines
may be used to further configure sets of features. Track lines
are placed at the beginning of the list of features they are
to affect.
Rules for sniffing as True::
- There must be 21 columns on each fields line
- matches, misMatches repMatches, nCount, qNumInsert,
qBaseInsert, tNumInsert, tBaseInsert, strand, qSize, qStart,
qEnd, tName, tSize, tStart, tEnd, blockCount, blockSizes,
qStarts, tStarts must be correct
- We will only check that up to the first 10 alignments are
correctly formatted.
>>> from galaxy.datatypes.sniff import get_test_fname
>>> fname = get_test_fname( '1.psl' )
>>> Psl().sniff( fname )
True
>>> fname = get_test_fname( '2.psl' )
>>> Psl().sniff( fname )
True
>>> fname = get_test_fname( 'interval.interval' )
>>> Psl().sniff( fname )
False
>>> fname = get_test_fname( '2.txt' )
>>> Psl().sniff( fname )
False
>>> fname = get_test_fname( 'test_tab2.tabular' )
>>> Psl().sniff( fname )
False
>>> fname = get_test_fname( 'mothur_datatypetest_true.mothur.ref.taxonomy' )
>>> Psl().sniff( fname )
False
"""
def check_items(s):
s_items = s.split(",")
for item in s_items:
if int(item) < 0:
raise Exception("Out of range")
count = 0
for line in file_prefix.line_iterator():
line = line.strip()
if not line:
break
if line:
if line.startswith("browser") or line.startswith("track"):
# Skip track lines.
continue
items = line.split("\t")
if len(items) != 21:
return False
# tName is a string
items.pop(13)
# qName is a string
items.pop(9)
# strand
if items.pop(8) not in ["-", "+", "+-", "-+"]:
raise Exception("Invalid strand")
# blockSizes
s = items.pop(15).rstrip(",")
check_items(s)
# qStarts
s = items.pop(15).rstrip(",")
check_items(s)
# tStarts
s = items.pop(15).rstrip(",")
check_items(s)
if any(int(item) < 0 for item in items):
raise Exception("Out of range")
count += 1
if count == 10:
break
if count > 0:
return True
lib/galaxy/datatypes/test/1.psl
+6
View File
@@ -0,0 +1,6 @@
browser position chr22:13073000-13074000
browser hide all
track name=fishBlats description="Fish BLAT" visibility=2 useScore=1
59 9 0 0 1 823 1 96 +- FS_CONTIG_48080_1 1955 171 1062 chr22 47748585 13073589 13073753 2 48,20, 171,1042, 34674832,34674976,
59 7 0 0 1 55 1 55 +- FS_CONTIG_26780_1 2825 2456 2577 chr22 47748585 13073626 13073747 2 21,45, 2456,2532, 34674838,34674914,
59 7 0 0 1 55 1 55 -+ FS_CONTIG_26780_1 2825 2455 2676 chr22 47748585 13073727 13073848 2 45,21, 249,349, 13073727,13073827
lib/galaxy/datatypes/test/2.psl
+5
View File
@@ -0,0 +1,5 @@
2890 41 0 0 25 41 25 32 + NZ_CP026740.1 98205 66320 69251 PS02056|SAMN29228906_5 3054 125 3047 51 72,57,4,9,13,13,86,4,44,4,19,2,1,52,7,8,4,181,108,189,50,41,3,231,2,3,19,3,5,61,136,1,8,30,18,1,42,26,55,119,2,380,3,109,5,566,2,2,45,11,34, 66320,66392,66449,66454,66464,66477,66490,66577,66581,66626,66630,66649,66653,66654,66706,66714,66723,66727,66908,67016,67206,67257,67298,67303,67540,67542,67545,67565,67569,67574,67635,67771,67774,67782,67813,67832,67834,67877,67903,67959,68079,68081,68462,68465,68575,68580,69154,69157,69159,69206,69217, 125,198,256,260,269,283,297,383,388,432,437,457,459,461,514,521,529,534,716,826,1015,1065,1108,1111,1342,1346,1351,1370,1373,1380,1442,1579,1580,1589,1619,1637,1638,1680,1708,1763,1882,1885,2265,2269,2378,2384,2950,2952,2955,3000,3013,
1513 16 0 0 12 16 14 20 - NZ_CP026740.1 98205 32103 33632 PS02056|SAMN29228906_6 1548 13 1546 27 94,6,159,153,17,2,74,7,6,28,147,6,1,7,202,4,3,164,1,1,2,276,3,83,4,21,42, 32103,32197,32203,32362,32515,32532,32535,32609,32618,32626,32654,32802,32808,32810,32818,33020,33026,33030,33194,33195,33197,33201,33477,33481,33565,33569,33590, 13,108,115,275,430,448,450,525,532,538,567,714,723,724,731,935,939,942,1107,1110,1111,1113,1390,1393,1476,1482,1504,
762 108 0 0 14 108 13 132 + NZ_CP026740.1 98205 69627 70497 PS02056|SAMN29228906_3 6215 1049 1943 28 81,1,45,4,35,2,66,1,2,99,41,2,5,4,40,76,3,3,72,4,3,2,7,71,5,11,3,74, 69627,69711,69713,69758,69762,69797,69799,69865,69868,69870,69969,70011,70014,70019,70023,70073,70150,70153,70157,70295,70304,70311,70314,70332,70404,70409,70420,70423, 1049,1130,1131,1177,1182,1218,1321,1388,1389,1392,1492,1533,1535,1543,1548,1588,1664,1669,1672,1744,1748,1751,1753,1760,1831,1837,1857,1869,
77 0 0 0 0 0 0 0 + NZ_CP026740.1 98205 70504 70581 PS02056|SAMN29228906_3 6215 134 211 1 77, 70504, 134,
2890 41 0 0 25 41 25 32 + CP026740.1 98205 66320 69251 PS02056|SAMN29228906_5 3054 125 3047 51 72,57,4,9,13,13,86,4,44,4,19,2,1,52,7,8,4,181,108,189,50,41,3,231,2,3,19,3,5,61,136,1,8,30,18,1,42,26,55,119,2,380,3,109,5,566,2,2,45,11,34, 66320,66392,66449,66454,66464,66477,66490,66577,66581,66626,66630,66649,66653,66654,66706,66714,66723,66727,66908,67016,67206,67257,67298,67303,67540,67542,67545,67565,67569,67574,67635,67771,67774,67782,67813,67832,67834,67877,67903,67959,68079,68081,68462,68465,68575,68580,69154,69157,69159,69206,69217, 125,198,256,260,269,283,297,383,388,432,437,457,459,461,514,521,529,534,716,826,1015,1065,1108,1111,1342,1346,1351,1370,1373,1380,1442,1579,1580,1589,1619,1637,1638,1680,1708,1763,1882,1885,2265,2269,2378,2384,2950,2952,2955,3000,3013,