[OMA] When input sequence data are DNA, OMA conversion in AA ?

Question

[pourbiostars]

Hello,

I'm actually using OMA with this parameter:

• " # Type of input sequence data, has to be either 'DNA' or 'AA' InputDataType := 'DNA'; "

So OMA knows that I'm using DNA sequences. But I find this parameter inside the parameter file:

• " # alignments which have a score lower than MinScore will not be considered. # The scores are in Gonnet PAM matrices units. MinScore := 181; "

Gonnet PAM matrices being amino acid based, I would like to know if OMA convert DNA into AA before the allVSall step ?

Thank you.

Best regards.

Answer 1

[Christophe Dessimoz]

No, OMA does not translate DNA into AA.

While it's true that Gonnet PAM matrices were defined on amino-acid, it's also possible to derive empirical scoring matrices from DNA in the same way. In OMA standalone, if the input data type is DNA, we use such matrices to perform alignment and evolutionary distance inference.

[pourbiostars]

Thank you for your answer,

When comparing species (from the same plant taxonomy family) does using CDS sequence instead of AA matter when the clustering is done ?

[Christophe Dessimoz]

We recommend using protein sequences, because we have done much more benchmarking on this type of data. There are practical advantages too: public genomes exported from OMA, including precomputed all-against-all, are only available at for protein sequences. By contrast, the DNA-level comparison support is still experimental in OMA. We have started using it for non-coding regions.

So for coding regions, I strongly recommend that you use the translated sequences.