Bwa Error While Trimming Color Space Data

Question

[Farhat]

I am using the following pipeline for aligning some paired end SOLiD data.

bwa aln -c -n 0.06 -o 2 -t 8 -q 10 ~/genomes/hydra/ACZUJGI/color/hydra ~/hydra/solid/hsamp_F3.fastq.gz > /scratch/hydra/hsamp_F3.sai
bwa aln -c -n 0.06 -o 2 -t 8 -q 10 ~/genomes/hydra/ACZUJGI/color/hydra ~/hydra/solid/hsamp_R3.fastq.gz > /scratch/hydra/hsamp_R3.sai

bwa sampe -P ~/genomes/hydra/ACZUJGI/color/hydra /scratch/hydra/hsamp_F3.sai /scratch/hydra/hsamp_R3.sai ~/hydra/solid/hsamp_F3.fastq.gz ~/hydra/solid/hsamp_R3.fastq.gz | samtools view -bS -|samtools sort - /scratch/hydra/hsamp_solid

On running this I get the following error. Just showing the last few lines from the output here. I only get this error when the -q parameter is nonzero.

[bwa_paired_sw] 91 out of 33101 Q17 discordant pairs are fixed.
[bwa_sai2sam_pe_core] time elapses: 74.36 sec
[bwa_sai2sam_pe_core] refine gapped alignments... 1.53 sec
[bwa_sai2sam_pe_core] print alignments... [samopen] SAM header is present: 20914 sequences.
Parse error at line 20916: sequence and quality are inconsistent

The error happens at the conversion to bam step in the pipeline. If I look at line 20916, it shows

1_29_54 141     *       0       0       *       *       0       0       NNCANGNAANANATCNNCCGGNTANANTTGANTTANNTTN        !!@;!9!:?!;!:>>!!8?57!66!7!8=<9!<<9!!-?!!!&!<!!!!!      XC:i:40

Thus the read is truncated but the quality line is not. Is there a workaround for this?

Answer 1

[brentp]

I think the proper answer is don't use -q with colorspace as it's designed for base-space. If you disregard that, you can pipe the output to this command (then to SAM)

$BWA_COMMAND \
| awk 'BEGIN{FS=OFS="\t"} \
      ($1 ~ /^@/){ print $0} \
      ($1 !~ /^@/){ $11 = substr($11, 0, length($10)); print $0}' \
| $SAMTOOLS_COMMAND > $OUT

That makes sure the qualities are the same length as the sequence.

You could also trim your reads with this: https://github.com/brentp/bio-playground/blob/master/solidstuff/solid-trimmer.py

[Farhat]

Thanks! I thought there may be some switch I am missing or something like that. The solid_trimmer looks useful!