A word about maq

[lh3]

I have not maintained maq for quite some time. Most of my time on sequencing related tools goes to samtools and bwa. I usually recommend people to use these tools (and other better tools) rather than maq. This does not mean maq is not accurate any more. On the contrary, several independent evaluations show that maq's alignment and SNP calling accuracy is still among the top few. What maq is missing is the speed of more recent aligners and the native SAM output.

That is saying: if you are happy with maq, you may continue to use it and do not worry maq is something substandard; if you need shining new features absent from maq or feel maq is too slow or maq2sam+fixmate is clumsy, try bwa/novoalign+samtools/picard/gatk/....

[aleferna]

time tested software

You know, I'm testing a few of the new aligner applications and I think maq and even blat are still quite relevant because they have been time tested. People have continued to use these applications even after new versions and fancy optimizations come out, trouble is. With all the complexity that new fancy algorithms bring into the mix its difficult to commit. You never really know if a new 1000x faster algorithm is going to fail miserably because a fraction of your reads have a particular characteristic that was not considered. New SW is buggy and a bug in the "newest" aligner might misdirect an experimenter to generate the wrong conclusions. With all the massive amounts of data generated in NGS its tough to catch these things. Specially now that sequencing is becoming the new microscope.

So, I say, embrace the new, but always keep the old time tested versions at hand and perform sanity checks.

[dawe]

Hi Heng,

several independent evaluations show that maq's alignment and SNP calling accuracy is still among the top few.

Is this still true?
d

[HESmith]

Hi Heng,

Am I correct in remembering that there is an upper limit for read coverage in MAQ?

Thanks,
Harold