Error in sample_phybreak: Missing Value in TRUE/FALSE Statement

[MarriaK]

Hi Don,
I encountered an error with the phybreak package: Error in if (runif(1) < pbe1$logLikseq - pbe0$logLikseq) : missing value where TRUE/FALSE needed. Could you please provide guidance on resolving this issue?

Best,
Maryam

[donkeyshot]

Hi MaryamI'm on holiday now and don't have a computer. I'm back in about two weeks time, sorry about that.But we can try. I've seen the question before and it probably has to do with the initial state and/or model choice, with the loglikelihood starting at minus infinity. After making the phybreak object with "something <- phybreak(...)", what is the value of "logLik(something)"?Best wishesDon Op 28 jul 2024 16:26 schreef Maryam Kafi Kang ***@***.***>: Hi Don, I encountered an error with the phybreak package: Error in if (runif(1) < pbe1$logLikseq - pbe0$logLikseq) : missing value where TRUE/FALSE needed. Could you please provide guidance on resolving this issue? Best, Maryam —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[MarriaK]

Thank you for your response. Here is the sample output before the model throws the error:

sample    logLik     mu  gen.mean  sam.mean parsimony (nSNPs = 3395)
0         -Inf    0.000103  1         1    111436

Have a great time!
Maryam

[donkeyshot]

Thanks. Yes, indeed the loglikelihood starts at minus infinity. I see there are many SNPs in the data, how many samples do you have? What pathogen is it? Are the SNPs reliable?Op 28 jul 2024 21:53 schreef Maryam Kafi Kang ***@***.***>: Thank you for your response. Here is the sample output before the model throws the error: sample logLik mu gen.mean sam.mean parsimony (nSNPs = 3395) 0 -Inf 0.000103 1 1 111436 Have a great time! Maryam —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***>

[MarriaK]

My data are generated using the FAVITES method, which is a simulation method. The dataset contains 560 samples.
Best,
Maryam

[donkeyshot]

Hi MaryamI'm afraid the simple answer then is that the dataset is too large. Maybe it is possible to construct an initial state that would get it started (you can provide a phylogenetic tree and transmission tree as initial state), but after that, phybreak will be extremely slow so I'm not sure if it's worth trying.If you really want to (I don't know what your research question is), you'll have to wait till I get home for me to be able to help.Sorry about that.Best wishes, DonOp 29 jul 2024 01:20 schreef Maryam Kafi Kang ***@***.***>: My data are generated using the FAVITES method, which is a simulation method. The dataset contains 560 samples. Best, Maryam —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***>