Alignpairedend: added alignment using shifting with best kmer similarity

(low level layer in C and Cython API)
This commit is contained in:
Celine Mercier
2019-01-21 17:30:46 +01:00
parent 59017c0d6b
commit c7f5b8d980
10 changed files with 934 additions and 89 deletions

View File

@ -0,0 +1,4 @@
#cython: language_level=3
cdef object buildAlignment(object direct, object reverse)

View File

@ -2,19 +2,20 @@
from obitools3.apps.progress cimport ProgressBar # @UnresolvedImport
from obitools3.dms import DMS
from obitools3.dms.view.view cimport View
from obitools3.dms.view import RollbackException
from obitools3.dms.view.typed_view.view_NUC_SEQS cimport View_NUC_SEQS
from obitools3.dms.column.column cimport Column, Column_line
from obitools3.dms.capi.obiview cimport QUALITY_COLUMN, COUNT_COLUMN, NUC_SEQUENCE_COLUMN, ID_COLUMN
from obitools3.dms.capi.obitypes cimport OBI_INT, OBI_STR, index_t, OBI_QUAL
from obitools3.dms.column.column cimport Column
from obitools3.dms.capi.obiview cimport QUALITY_COLUMN
from obitools3.dms.capi.obitypes cimport OBI_QUAL
from obitools3.apps.optiongroups import addMinimalInputOption, addMinimalOutputOption
from obitools3.uri.decode import open_uri
from obitools3.apps.config import logger
from obitools3.libalign._qsassemble import QSolexaReverseAssemble
from obitools3.libalign._qsrassemble import QSolexaRightReverseAssemble
from obitools3.libalign._solexapairend import buildConsensus, buildJoinedSequence
from obitools3.dms.obiseq cimport Nuc_Seq_Stored, Nuc_Seq
from obitools3.dms.obiseq cimport Nuc_Seq
from obitools3.libalign.shifted_ali cimport Kmer_similarity, Ali_shifted
import sys
import os
@ -27,6 +28,7 @@ REVERSE_QUALITY_COLUMN_NAME = b"REVERSE_QUALITY"
__title__="Aligns paired-ended reads"
def addOptions(parser):
addMinimalInputOption(parser)
@ -41,14 +43,6 @@ def addOptions(parser):
type=str,
help="URI to the reverse reads if they are in a different view than the forward reads")
# TODO
# group.add_argument('--index-file',
# action="store", dest="alignpairedend:index",
# metavar="<FILENAME>",
# default=None,
# type=str,
# help="URI to the index reads")
group.add_argument('--score-min',
action="store", dest="alignpairedend:smin",
metavar="#.###",
@ -56,12 +50,22 @@ def addOptions(parser):
type=float,
help="Minimum score for keeping alignments")
group.add_argument('-A', '--true-ali',
action="store_true", dest="alignpairedend:trueali",
default=False,
help="Performs gap free end alignment of sequences instead of using kmers to compute alignments (slower).")
group.add_argument('-k', '--kmer-size',
action="store", dest="alignpairedend:kmersize",
metavar="#",
default=3,
type=int,
help="K-mer size for kmer comparisons, between 1 and 4 (not when using -A option; default: 3)")
# TODO declarations, cdef, imports
la = QSolexaReverseAssemble()
ra = QSolexaRightReverseAssemble()
def buildAlignment(object direct, object reverse):
cdef object buildAlignment(object direct, object reverse):
if len(direct)==0 or len(reverse)==0:
return None
@ -82,9 +86,9 @@ def buildAlignment(object direct, object reverse):
ali = rali
return ali
def alignmentIterator(entries):
def alignmentIterator(entries, aligner):
if type(entries) == list:
two_views = True
@ -94,7 +98,7 @@ def alignmentIterator(entries):
else:
two_views = False
entries_len = len(entries)
for i in range(entries_len):
if two_views:
seqF = forward[i]
@ -104,9 +108,12 @@ def alignmentIterator(entries):
seqR = Nuc_Seq(seqF.id, seqF[REVERSE_SEQ_COLUMN_NAME], quality=seqF[REVERSE_QUALITY_COLUMN_NAME])
seqF.pop(REVERSE_SEQ_COLUMN_NAME)
seqF.pop(REVERSE_QUALITY_COLUMN_NAME)
ali = buildAlignment(seqF, seqR)
ali = aligner(seqF, seqR)
if ali is None:
continue
yield ali
@ -160,11 +167,6 @@ def run(config):
else:
entries_len = len(entries)
# if "index" in config["alignpairedend"]: # TODO
# index = open_uri(config["alignpairedend"]["index"])
# else:
# index = None
# Open the output
output = open_uri(config['obi']['outputURI'],
input=False,
@ -174,22 +176,30 @@ def run(config):
view = output[1]
Column.new_column(view, b"QUALITY", OBI_QUAL) #TODO output URI quality option
Column.new_column(view, QUALITY_COLUMN, OBI_QUAL) #TODO output URI quality option?
if 'smin' in config['alignpairedend']:
config['alignpairedend']['sminL'] = config['alignpairedend']['smin']
config['alignpairedend']['sminR'] = config['alignpairedend']['smin']
sminL = config['alignpairedend']['sminL']
sminR = config['alignpairedend']['sminR']
smin = config['alignpairedend']['smin']
else:
sminL = 0
sminR = 0
smin = 0
# Initialize the progress bar
pb = ProgressBar(entries_len, config, seconde=5)
ba = alignmentIterator(entries)
if config['alignpairedend']['trueali']:
kmer_ali = False
aligner = buildAlignment
else :
kmer_ali = True
if type(entries) == list:
forward = entries[0]
reverse = entries[1]
aligner = Kmer_similarity(forward, view2=reverse, kmer_size=config['alignpairedend']['kmersize'])
else:
aligner = Kmer_similarity(entries, kmer_size=config['alignpairedend']['kmersize'])
ba = alignmentIterator(entries, aligner)
i = 0
for ali in ba:
@ -197,31 +207,33 @@ def run(config):
consensus = view[i]
if sminL > 0:
if ((ali.direction=='left' and ali.score > sminL)
or (ali.score > sminR)):
buildConsensus(ali, consensus)
if not two_views:
seqF = entries[i]
else:
seqF = forward[i]
if smin > 0:
if (ali.score > smin) :
buildConsensus(ali, consensus, seqF)
else:
seqF = ali[0].wrapped
if not two_views:
seq = entries[i]
seqR = Nuc_Seq(seq.id, seq[REVERSE_SEQ_COLUMN_NAME], quality = seq[REVERSE_QUALITY_COLUMN_NAME])
seqR = Nuc_Seq(seqF.id, seqF[REVERSE_SEQ_COLUMN_NAME], quality = seqF[REVERSE_QUALITY_COLUMN_NAME])
else:
seqR = reverse[i]
buildJoinedSequence(ali, seqR, consensus)
buildJoinedSequence(ali, seqR, consensus, forward=seqF)
consensus[b"sminL"] = sminL
consensus[b"sminR"] = sminR
consensus[b"smin"] = smin
else:
buildConsensus(ali, consensus)
# TODO
# if "index" in config['alignpairedend'] and config['alignpairedend']['index'] is not None:
# idx = str(config['alignpairedend']['index'].next())
# consensus[b"illumina_index"] = idx
buildConsensus(ali, consensus, seqF)
if kmer_ali :
ali.free()
i+=1
if kmer_ali :
aligner.free()
# Save command config in View and DMS comments
command_line = " ".join(sys.argv[1:])
view.write_config(config, "alignpairedend", command_line, input_dms_name=input_dms_name, input_view_name=input_view_name)