Implementing more protein models into BEAUTI/BEAST

[JirkaMP]

Hi everyone,
I'm very new to the BEAST community and tumbled across an issue.
Using IQtree's ModelFinder, I identified the best protein substitution models for a set of proteins I included in a phylogenetic analysis. Now, I would like to run a molecular dating analysis with BEAST including those protein models, so I created a nexus file containing my multiprotein alignment, regions that refer to each protein and their respective best models (see below):

##The alignment was too long to add##

begin sets;
charset 38k = 1-699;
charset ac81 = 700-1009;
charset dnapol = 1010-3374;
charset helicase = 3375-6193;
charset lef4 = 6194-7531;
charset lef8 = 7532-9533;
charset lef9 = 9534-10914;
charset p33 = 10915-11677;
charset p47 = 11678-12851;
charset p74 = 12852-13939;
charset pif1 = 13940-14875;
charset pif2 = 14876-15459;
charset pif3 = 15460-15813;
charset pif4 = 15814-16667;
charset pif6 = 16668-17071;
charset vp39 = 17072-17615;
charset vp91 = 17616-19070;
end;

begin assumptions;
charset_model 38k = Q.pfam+F+I{0.0147265}+G4{1.30949};
charset_model ac81 = LG+F+I{0.0308603}+G4{1.61915};
charset_model dnapol = Q.pfam+F+I{0.0125308}+G4{1.47578};
charset_model helicase = Q.pfam+F+I{0.0034635}+G4{2.03655};
charset_model lef4 = Q.pfam+F+I{0.0219115}+G4{1.69875};
charset_model lef8 = Q.pfam+F+I{0.00791903}+G4{1.34902};
charset_model lef9 = VT+F+I{0.00778775}+G4{1.9748};
charset_model p33 = VT+F+I{0.00970993}+G4{2.68922};
charset_model p47 = Q.pfam+F+I{0.00839166}+G4{1.56855};
charset_model p74 = Q.pfam+F+I{0.0200284}+G4{1.70337};
charset_model pif1 = WAG+F+I{0.0197432}+G4{2.26165};
charset_model pif2 = WAG+F+I{0.0276439}+G4{1.4145};
charset_model pif3 = Q.yeast+I{0.0323176}+G4{1.1696};
charset_model pif4 = Q.pfam+F+I{0.0101419}+G4{1.25143};
charset_model pif6 = Q.yeast+G4{2.02776};
charset_model vp39 = LG+F+I{0.0149826}+G4{2.2551};
charset_model vp91 = WAG+F+I{0.0100864}+G4{2.99185};
end;

Importing the file into BEAUTI works with no issue, but I noticed that only the WAG model is available for the models that I would need, but all the others aren't available.
My question, is there a way to implement these missing models (like VT or LG) or maybe a command line based approach to circumvent this restriction of BEAUTI?

Thank you for taking the time to read my post!
Please let me know if I described my issue too superficially and if you require more information.

Best
Jirka

[GuyBaele]

Hi Jirka,

While I don't think that all of these models are currently implemented in BEAST, some of them certainly are: BLOSUM_62, DAYHOFF, JTT, MT_REV_24, CP_REV_45, WAG, LG, FLU, MTVER, MTPRO, MTMET, MTINV, MTDEU and MTMAM.

I'm not sure why (some of) these models are not showing up in BEAUti but can certainly look into it. Could you e-mail me your input file in the meantime?

Best regards,
Guy

[JirkaMP]

Dear Guy, Thank you for your rapid response! Please find attached the nexus file I use as input and a screenshot from my Beauti showing the available models I can choose from. Cheers Jirka

…

________________________________ Von: Guy Baele ***@***.***> Gesendet: Dienstag, 28. Mai 2024 15:08 An: beast-dev/beast-mcmc ***@***.***> Cc: Petersen, Jirka Manuel ***@***.***>; Author ***@***.***> Betreff: Re: [beast-dev/beast-mcmc] Implementing more protein models into BEAUTI/BEAST (Issue #1184) Hi Jirka, While I don't think that all of these models are currently implemented in BEAST, some of them certainly are: BLOSUM_62, DAYHOFF, JTT, MT_REV_24, CP_REV_45, WAG, LG, FLU, MTVER, MTPRO, MTMET, MTINV, MTDEU and MTMAM. I'm not sure why (some of) these models are not showing up in BEAUti but can certainly look into it. Could you e-mail me your input file in the meantime? Best regards, Guy — Reply to this email directly, view it on GitHub<#1184 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BHTSFTUYD3VNCQ73RKA4MEDZER6VJAVCNFSM6AAAAABIM4EXRKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMZVGE3TOMJWHA>. You are receiving this because you authored the thread.Message ID: ***@***.***>

[GuyBaele]

Hi Jirka,

I see all of those models that I mentioned when I run BEAUti, so I currently can't replicate what you're experiencing. Let's for the time being move this issue off of GitHub. Please contact me via e-mail - not by responding to this GitHub notification - so that we can look into this.

Best regards,
Guy