bwa error: paired reads have different names

Question

[vjmorley]

I'm getting an error when I try to pair my reads using bwa (version 0.7.10-r789). Here's my pipeline:

Trim for quality using fastx fastq_quality_trimmer

fastq_quality_trimmer -t 20 -l 30 -Q33 -i for.fastq -o for_trimmed.fastq
fastq_quality_trimmer -t 20 -l 30 -Q33 -i rev.fastq -o rev_trimmed.fastq

Align forward and reverse reads to reference

bwa index ref.fa
bwa aln ref.fa for_trimmed.fastq > for.sai
bwa aln ref.fa rev_trimmed.fastq > rev.sai

Pair reads, for which I get the error in the following output:

bwa sampe for.sai rev.sai for_trimmed.fastq rev_trimmed.fastq > sample.sam

[bwa_sai2sam_pe_core] convert to sequence coordinate...
[infer_isize] fail to infer insert size: too few good pairs
[bwa_sai2sam_pe_core] time elapses: 0.14 sec
[bwa_sai2sam_pe_core] changing coordinates of 0 alignments.
[bwa_sai2sam_pe_core] align unmapped mate...
[bwa_sai2sam_pe_core] time elapses: 0.00 sec
[bwa_sai2sam_pe_core] refine gapped alignments... 0.07 sec
[bwa_sai2sam_pe_core] print alignments... [bwa_sai2sam_pe_core] paired reads have different names: "HWI-D00306:565:C6NPTANXX:8:1101:9745:2553", "HWI-D00306:565:C6NPTANXX:8:1101:9672:2717"

Any suggestions on how to fix this problem or remove the offending reads?

Answer 1

[User 59]

Because you're trimming your reads independently you're ensuring that their order is out of sync, bwa relies on your files being synchronised in terms of read names. Use a trimmer that can maintain the order of your fastq files by throwing out both pairs of reads if one of them fails a quality check.

Cutadapt can do paired-end aware quality and adaptor trimming.

https://cutadapt.readthedocs.org/en/stable/guide.html#trimming-paired-end-reads

[vjmorley]

Thanks! I used cutadapt to trim the paired reads, and that solved the problem. I really appreciate the help!

[Ashutosh Pandey]

As Daniel suggested that you should retrim the fastq reads using a trimmer that maintains the order of reads in the two fastq files. Trimmomatic is another paired-end aware trimmer. Alternatively, you can remove orphan reads as described here: Combining The Paired Reads From Illumina Run