Casava 1.8 Fastq format for using with BWA

[a14418e10]

Hello,

I'm mapping paired-end reads coming from Illumina HiScanSQ using BWA
and I'm wondering if BWA correctly recognizes the Fastq sequence identifier line as
used by Casava 1.8.

For instance, if I have the following line in my Fastq
@EAS139:136:FC706VJ:2:2104:15343:197393 1:Y:18:ATCACG
do I have to change it to, say,
@EAS139:136:FC706VJ:2:2104:15343:197393#1
so that BWA knows that this read is the first one in a read pair.

The version of BWA is bwa-0.5.9rc1

Thanks in advance !

[BaCh]

@EAS139:136:FC706VJ:2:2104:15343:197393#1

Make that

@EAS139:136:FC706VJ:2:2104:15343:197393/1

See also http://en.wikipedia.org/wiki/FASTQ_f...ce_identifiers

B

[a14418e10]

Thanks !

Is it crucial to keep the "#0" ( or "#ATCACG" ) part in the transformed sequence ?

[swbarnes2]

Bwa doesn't know that it's read 1 because of the name. It's not that smart. You give it two fastq files, and the first is all the read 1 reads, and the second is the read 2 reads, and it assumes that the first reads of each is a pair, and the second read of each is a pair, etc. If that assumption is wrong, it's not going to read the names and figure out how things should really be paired. It's going to go with that assumption, and give you very odd pairs.