Excluding sites

[fluttercork]

Would it be possible to have raxml-ng recognize site weights of 0 (or, alternatively, to implement the exclude file of RAxML?), which would be a quick and easy way for the user to exclude sites from the analysis?

When working with MSAs with multiple partitions, I often want to exclude one or more partitions, which could be more easily done using a weight / exclusion file than rebuilding the MSA each and every time.

Thanks,

Olaf

[liamxg]

Dear @fluttercork,
Good idea!

[amkozlov]

Dear Olaf,

thanks for the suggestion, I will put in on my list

[amkozlov]

Site exclusion is now supported in the dev branch in the following way:

1. In partition file, individual partitions can be commented out, e.g.

 LG+G, p1 = 1-20, 81-113
 #disable partition p2
 #WAG+G, p2 = 21-50
 JTT, p3 = 51-70

2. To prevent excluding MSA sites by mistake, you need to add --force msa_extra_cols to explicitly disable partition completeness check:

$ raxml-ng --msa prot21.fa --model prot21.excl.part  --force msa_extra_cols

...

[00:00:00] Reading alignment from file: /home/alex/hits/ngtest/data/prot21.fa
[00:00:00] Loaded alignment with 21 taxa and 113 sites

WARNING: Found 40 alignment site(s) which are not assigned to any partition. They will be ignored.

Alignment comprises 2 partitions and 73 patterns

Partition 0: p1
Model: LG+G4m+B
Alignment sites / patterns: 53 / 53
Gaps: 13.93 %
Invariant sites: 13.21 %

Partition 1: p3
Model: JTT+B
Alignment sites / patterns: 20 / 20
Gaps: 1.90 %
Invariant sites: 15.00 %