Supertree With Matrix Representation With Parsimony

Question

[Pavid]

Hi!

I'm working with tuberculosis data, we have SNPs, MIRUs, RFLP and Spoligotype data.
I would like to do a supertree (phylogenetic approach in which many overlapping trees are combined to produce a single tree) combining these four methods.

I've read some papers and I've notice the Matrix Representation with Parsimony is very used.

I have about 1500 samples for 90 SNPs. I've done an individual tree for SNPs with Biopython and Networkx.

answering Leszek comment:
the output of MIRU (Mycobacterial Interspersed Repetitive Units) are numbers (1 to 25)
the output of Spoligotype (0 and 1 - presence and absence)

Does anyone has done something like that?

Or has any direction?

Any help is welcome.

Thanks :)

[Leszek]

How many samples do you have? And how many SNPs per sample?

Answer 1

[Leszek]

I've never done it myself, but after discussion with my friend, I think you could simple concatenate all your SNP positions into multiple fasta file and try to use any tree reconstruction program (f.e.: phyml or raxml) to get tree based on SNPs.

I don't know what MIRUs and Spoligotype data are...

[Pavid]

I've updated my question with the output of MIRU and spoligotype. The problem is not the SNP tree. Is how to concatenate

Answer 2

[DG]

Personally I'm not a fan of Supertree methods. I tend to thing the supermatrix approach (concatenate data together), partition where appropriate, and doing normal ML reconstruction on the data works better. Note that programs like RAxML and a few others can handle things like continuous characters, etc.

[Pavid]

what I want with this is to do a tree that would concatenate several trees from different methods together. Doesn't have to be with the MRP method, if you have another solution

[DG]

RAxML will do what you want to do. You can concatenate your disparate data together, specify the individual data partitions (SNP, MIRU, RFLP, and Spoligotype), appropriate things like ML models of substitution, and reconstruct the ML tree. Although the problem may be coming up with appropriate models for the non-character data.

[Pavid]

"RAxML is a fast implementation of maximum-likelihood (ML) phylogeny estimation that operates on both nucleotide and protein sequence alignments" will this be able to use numeric data? output of MIRU are numbers 1 to 25 and spoligotype 0 and 1.

[DG]

newer versions of RAxML have also implemented binary and multi-state models of evolution. The Multi-State option allows for up to 32 states. The Multi-State option allows for the MK, Ordered, or GTR model of evolution (GTR is the default).

See: http://wwwkramer.in.tum.de/exelixis/software.html

[Pavid]

ok, so I'm having some difficulties in testing this software. I've done a test file with dna and binary, but gives me a error into the partitions. But I can make another question with more details

[Pavid]

ok, I'm having some problems in getting an example file to work. I'm gone make another question to understand how this work