[bwa] generating paired end sam file from sai files

[insular]

Hi all,

I was aligning paired end reads in two fastq files: read_1.fq and read_2.fq generated from the grinder simulation program and kept the alignments in read_1.sai and read_2.sai files. Then I tried to combine them into one bam file by:

bwa sampe ../../CP000948.fna read_1.sai read_2.sai read_1.fq read_2.fq >read_12.sam

It will be a success when read_1.fq and read_2.fq are all properly paired (every read in 1 has a mate in 2) but it will report following error when I artificially remove one read from 1 (making one pair missing an end):

[bwa_sai2sam_pe_core] convert to sequence coordinate...
[fread] Unexpected end of file

My purpose of doing this is the fq files generated from grinder behave this way. Although their paper don't touch on this, I guess its one of the possible error mechansims they try to simulate. But this is causing trouble in the downstream bwa pipeline. Just wonder if this is a normal bwa sampe behavior and is there any mature tools out there for taking care of this. It is probably better than I manually write a more error-prone script to fix it myself.

Thanks a lot!

[krobison]

Yes, the convention is that your files must have properly paired reads & bwa is assuming your file is corrupted if that is not the case.

There is a good need for a tool to fix this, though actually implementing it isn't trivial (in particular, what assumptions & memory requirements would such a tool have?).