BWA - sampe max occurrences of a read for pairing

[madsaan]

Hello,

Can anyone explain what the parameter -o means for sampe.According to BWA manual it says "A read with more occurrences will be treated as a single-end read.Reducing this parameter helps faster pairing".

Does it mean uneven number of reads in the two input fastq files?How do you decide what parameter to use?

Any help will be greatly appreciated.

Thanks,
Joji

[dp05yk]

BWA determines the suffix array intervals for each read during the 'aln' stage. If you're allowing errors (bwa aln -n #oferrors), a read can have many suffix array intervals of varying lengths.

The more (and larger) suffix array intervals a read has, the more work BWA has to go through during 'sampe' to pair it with its mate as there is a greater combinations of positions to consider for each mate... if the number of potential alignments exceeds the -o parameter (defaulted at 100 000), BWA just skips pairing altogether and leaves each mate aligned as if they were not mates at all in order to maintain an efficient alignment process.