FamSeq vs Polymutt vs DeNovoGear

[Kath]

Hi,

I would like to call variants jointly in a trio, taking pedigree info into account. I have heard of three tools for doing this - FamSeq (http://odin.mdacc.tmc.edu/~wwang7/FamSeqIndex.html), PolyMutt (http://genome.sph.umich.edu/wiki/Polymutt) and DeNovoGear (http://sourceforge.net/projects/denovogear/).

Has anyone used any of these and can they advise me on what are the advantages and disadvantages of each?

Has anyone used any other tools for this?

Thanks!

Kath

[Elsie]

I've used FamSeq (piece of cake to use and helpful authors) and PolyMutt (again easy to use and helpful authors) but not DeNovoGear. FamSeq mainly just provides probabilities whereas PolyMutt does a bit more, more active filtering wherease Famseq provides the probabilities and expects you to filter. I have heard good things about DeNovoGear but never used it. The two I've used are fast and easy, give them a go and look at your results.

[neio1114]

I've used FamSeq (piece of cake to use and helpful authors) and PolyMutt (again easy to use and helpful authors) but not DeNovoGear. FamSeq mainly just provides probabilities whereas PolyMutt does a bit more, more active filtering wherease Famseq provides the probabilities and expects you to filter. I have heard good things about DeNovoGear but never used it. The two I've used are fast and easy, give them a go and look at your results.

I was trying to use polyMutt also, but could not understand how to generate .dat and .gif files.
can you explain in detail how to get those 2 file format??

[neio1114]

I am also trying to use PolyMutt. but I cannot understand how to generate .dat and .glf file. can you explain how to generate those files in detail?
Thank you

[mmiossec]

Has anyone used any other tools for this?

I would be interested to see an answer to this question. Ideally a tool that doesn't require another variant caller (SAMtools or GATK) to provide initial genotype predictions (as FamSeq and, to an extent, Polymutt, do).

Matt

[Len Trigg]

While this post is a little old, it is worth noting that the joint Bayesian variant calling of RTG Core does directly incorporate pedigree information during the primary variant calling, right where it can have the most effect.

Pedigree post-processors (such as FamSeq, Polymutt, DeNovoGear) are probably fine for SNPs, but for more complex variants (where the hypothesis space is much larger) the caller has already made a pedigree-oblivious decision as to which potential alleles to output GLs for.