the number of aligned reads by maq and bwa

[totalnew]

We have applied maq and bwa to align the same production data, and we see an obvious difference on the number of aligned reads between maq and bwa, especially a decrease in the number with longer read length.

This issue is critical, is it a drawback of bwa that aligns much less reads than maq? I am using the default options for bwa running, anyone has experience to increase the aligned reads dramatically by using different bwa options?

I don't know if it is a bottleneck of BWT algorithm even it is much faster than maq. If so, I might have to think again if I am really in favor of bwa. Now bwa is pushed forward, anyone has comments of this issue?

thanks

[joa_ds]

have you analysed your data further downstream ?

you are better off with 40% mapped sequences and good results than with 60% mapped rubbish.

what is the type of experiment you are doing at the moment? You might consider tweaking the options a bit in function of your experiment. If you are mRNA expression counting, you might allow lower quality reads to be mapped because both the control and sample of interest will have the same amount of 'rubbish match'. When you are SNP hunting you might use more stringent parameters to get rid of false positives etc.

You can always set filters to have almost everything mapped, but I do not think that is the way you want to go?

[bioinfosm]

BWA even allows gaps, so essentially more reads should be mapped than those from gapless MAQ alignment.

[lh3]

maq is more tolerant with the mismatches toward the 3'-end. Given typical sequencing error rate, the mapping ratio of maq and bwa is quite similar to each other. However, when the sequencing error rate is high, maq maps more reads. At the same time, one may consider to run bwa with "bwa aln -q15" to enable base trimming. This only works when the base quality is not rubbish.