+1 for manta and that long pipe :-D
• 1 views
•
link
written
by
WouterDeCoster
Help with .sam files _1,_2 from bowtie2 output into svDetect
I am trying to run a PE library through bowtie2 and then into svDetect for Circos.
Bowtie2 run with
module load Bowtie2/2.3.4.1-foss-2016b
bowtie2 -x w82 -1 SRR2163311_1.fastq.gz -2 SRR2163311_2.fastq.gz --un-conc unPEK.sam --al-conc alPEK.sam --met-file metPEK.txt -S samPEK.sam
My output provides me with alPEK.1.sam, alPEK.2.sam, samPEK.sam as well as in .txt.
Trying svDETECT with .conf file as (have tried using a comma to list both on the alPEK.sams)
general
input_format = sam
sv_type = all
mates_orientation=FR
read1_length=90
read2_length=90
mates_file=alPEK.sam
cmap_file=gm20.len
output_dir=results
tmp_dir=tmp
num_threads=14
general
detection
split_mate_file=1
window_size=750 step_length=150
/detection
filtering
split_link_file=0
strand_filtering=1
Output with error
# Linking procedure ...
-- file=/scratch/sxd/workDir/svDETbase/alPEK.sam
-- input format=sam
-- type=all
-- read1length=90, read2length=90
-- window size=750, step length=150 /usr/local/apps/eb/SVDetect/0.7-foss-2016b-Perl-5.24.1/bin/SVDetect:
can't open /scratch/sxd/workDir/svDETbase/alPEK.sam:No such file or
directory
Is it possible to input both .1.sam & .2.sam? I did try view/merge via samtools but unsuccessful. Is the bowtie2 output salvageable for svDetect or should I start over with BWA?
thanks!
1 answer
Bowtie2 does not support split reads, which most SV callers depend on. Use BWA mem, which is more suited for this. I am not familiar with SVdetect, but it seems quiet old. You might consider more recent alternatives like Manta from Illumina or Lumpy from the Hall/Quinlan-labs (list not comprehensive of course).
Also, there is no need to store unsorted SAM files. Better produce sorted BAM files with duplicates marked right away.
I recommend:
bwa mem (options...) | samblaster --ignoreUnmated | sambamba view -f bam -S -l 1 -o /dev/stdout /dev/stdin | sambamba sort --tmpdir=./ -o sorted.bam /dev/stdin
This gives a duplicate-marked and sorted BAM file, with an index produced on the fly and without any intermediate files that only consume space.
IF you really want to use bowtie2, The GRIDSS SV caller includes a gridss.SoftClipsToSplitReads utility program that will feed the soft clipped portion of the reads back to an aligner of your choice and convert them to split reads. Even works on the output of bwa which, somewhat surprisingly, results in bwa identifying split reads from reads that bwa only reports a soft clipped reads when given the entire read to align.
Log in to answer this question.
Thank you! I will opt for bwa mem then. I also appreciate the newer alternatives. I will get this running but thank you very much!
Please use
ADD COMMENT/ADD REPLYwhen responding to existing answers/comments to keep threads logically organized.If an answer was helpful you should upvote it, if the answer resolved your question you should mark it as accepted.
