How exactly does MAQ handle mate-pair data

Question

[muraved]

Hi,

I'm trying to align a 3.5K-insert mate-pair library using MAQ 0.7.0. From the man page (http://maq.sourceforge.net/maq-manpage.shtml), it is not quite clear to me how exactly MAQ handles the outward orientation of such reads, and how it distinguished paired-end input from mate-pair input. Do I have to pass the files in reverse order, i.e. R.fq F.fq? Do I have to reverse the sequences within the files? Both? Nothing?

Any help is greatly appreciated :-)

[GenoMax]

Is there a specific reason you are using this relatively ancient aligner?

[muraved]

Yes. The way it does consensus calling lends itself to read correction. I'm not so much interested in the alignment itself. Also, old doesn't mean bad ;-)

[GenoMax]

I never said MAQ was bad :) I have not used MAQ but perhaps suggestions in this thread may help (Aligning Mate Pair Data )

[muraved]

Thanks for the link. I am well aware of several read mappers for MP mapping, I'm really asking specifically about MAQ. I do appreciate your help, though :-)

Answer 1

[h.mon]

The MAQ man page has all the info you want:

-a INT Maximum outer distance for a correct read pair [250]

-A INT Maximum outer distance of two RF paied read (0 for disable) [0]

NOTE: Currently, read pairs from Illumina/Solexa long-insert library have RF read orientation. The maximum insert size is set by option -A. However, long-insert library is also mixed with a small fraction of short-insert read pairs. -a should also be set correctly.

You supply the reads as is, MAQ will guess by orientation and insert-size if they are paired-reads or mate-pairs.

[muraved]

As you will notice, my question has the exact same link you provided. The specifics I am asking for are not explained on the man page in my opinion. But maybe I'm overlooking something.

[h.mon]

Speaking about the manual, did you see this:

Maq cannot align reads longer than 63bp.

Reads this short are rare nowadays, do your reads conform to this length?

[muraved]

Oops... thanks for pointing that out, I must've missed that :-/

[h.mon]

Do I have to pass the files in reverse order, i.e. R.fq F.fq? Do I have to reverse the sequences within the files? Both? Nothing?

In my understanding, you should do nothing, but set -a and -A to reasonable values to your library.

Or you could pre-process you mate-pair files with NxTrim, keep -A 0, and set -a 5000 (or a reasonable maximum, depending on the skills of who prepared the mate-pair libraries).

Or you could pre-process your mate-pairs with NxTrim using the --rf flag, and then set -a 0 -A 5000 for MAQ.

If in doubt, create a small test case, with known proportions of paired-end and mate-pair reads, map the files and examine the results - this should settle your doubts.

Answer 2

[muraved]

Thanks. Hmm, so, there is no input specification of whether it's paired-end or mate-pair. For an input like maq map F.fq R.fq, it simpy checks the relative positions of mapping locations, and accepts F-R as specified by -a, and R-F as specified by -A then?