Mosaik alignment

[sdriscoll]

Dear All,

I am using MosaikAligner with -m all option. This option map a read into all possible locations (each location with a certain alignment quality). Is there a way to get the alignment with the highest score and to discard the rest of the alignments?

Thanks!!!

Ernesto

I think most of the time the "highest scoring alignment" is going to be a little ambiguous. If you're aligning short reads ( < 50 bp ) then you're gonna get reads that align in lots of places with the same number of mismatches and virtually no way to say that one is better than another.

I'm not sure what's out there that could do this for you but what might work is to let Mosaik just go for it (-m all) and then to do some post sorting on the output alignment file by binning read IDs that show up in multiple rows and sort their alignment scores. I still think what you'll end up with is a list of alignment locations that have the same score instead of single locations with the highest score as you're looking for. You'd want to look at # of mismatches as well as the alignment scores.

[donniemarco]

same error

hi everyone, i got the same error:
***************
hi all I m using Mosaik and create jump database with Mosaikjumb command which is successfully create the jump database but after that when i run MosaikAligner command it give error like this ERROR: A position (1099511627520) was specified that is larger than the jump positions database (113492972).
***********

Has anyone got an idea about it. I have submitted this error with mosaik error tracking too.
My parameters are:
-hs 15 -mmp .05 -m all -mhp 100 -act 26 -p 8 -j jumpdb.

Thanks!!!

[shahid.manzoor]

hi all I m using Mosaik and create jump database with Mosaikjumb command which is successfully create the jump database but after that when i run MosaikAligner command it give error like this ERROR: A position (1099511627520) was specified that is larger than the jump positions database (113492972).
So any body can help me to about this error.

[ernesto]

Thanks for you replies, but as far as I understand the -rmm option in MosaikSort is for paired-end reads, and my data is single ended.
Do you think this option will work with single-end reads as well?

Thanks,

Ernesto

[amcrisan]

As I understand it, the non-unique reads will not necessarily be discarded. They will be placed asides and may be resolved at later steps using Mosaik sort (the -rmm option). If you don't choose to do this, then I think the reads are essentially discarded.

[zhanxw]

From the documentation, -m unique, only place uniquely mapped read.
Does it mean Mosaik will discard the reads that can aligned at multiple locations?

[amcrisan]

That's difficult to say. The -m unique option will avoid giving you all the possible reads like the -m all option. However, all it does (according to the documentation) is really set aside any reads which can map to multiple locations, it doesn't take the highest scoring alignment out of all the possible alignment locations for that read. Additionally I don't see any indication that it actually would.

The best you can do in my opinion is to use the -m unique option, and then later you can use MosaikSort and tell it to try resolve multiple vs multiple (-rmm) read pairs, and so on. That being said, you may still get things discarded, as it may not be possible to resolve the multiple vs. multiple.

Lastly, using the -m unique option is faster and takes less space. You could also use the -m all option followed by MosaikSort for I think similar outcomes to using the unique option (not 100% sure though..).

Hope that's helpful.

[ernesto]

Dear All,

I am using MosaikAligner with -m all option. This option map a read into all possible locations (each location with a certain alignment quality). Is there a way to get the alignment with the highest score and to discard the rest of the alignments?

Thanks!!!

Ernesto