search

spond / HBL

Various HyPhy Batch Language Files
1 stars 3 forks source link

'nan' value in the result file #1

Open alborznoorani opened 8 years ago

alborznoorani commented 8 years ago

Hi everyone, I’m trying to calculate dN/dS values with FUBAR. In the .fubar.csv result file, all values for each codon are “nan”. As I checked, all values from phase 3 output (.samples.C) are also “nan”. I was wondering what the problem could be and how I can fix it. Thank you so much for helping me, Alborz

spond commented 8 years ago

Dear @alborznoorani,

Can you please provide the data file and run settings used to generate this error for me to reproduce and debug the error. Is it reliably reproducible from run to run? Does it occur only for one input alignment, or several different alignments?

Sergei

PS Have you tried the FUBAR.bf file included in the HyPhy distribution? See https://github.com/veg/hyphy-tutorials/tree/master/selection#use-fubar-to-find-sites-which-have-experienced-pervasive-diversification

alborznoorani commented 8 years ago

@spond Dear Sergei, Thank you very much for your attention. I’ve included the aligned fasta file, its corresponding tree file and the message log from one of my trials in this post. The issue has occurred for a number of different alignments and it is reproducible from run to run. I originally ran the program with all default setting and values but I have also tried with different setting values for the MCMC chain, its length and burn-in and … but I got the same error. I also ran the data with the FUBAR included in the HyPhy distribution but got the same result. Thanks again for your help, Alborz

messages.log.txt tree.txt aligned_fasta.txt

spond commented 8 years ago

Dear @alborznoorani,

I've tracked the issue down to an obscure internal HyPhy bug that has to do with string pattern matching. A quick workaround is to rename the sequences that start with mu to start with something else (anything will do). Give that a go and see if the code results make sense now.

Sergei

alborznoorani commented 8 years ago

Dear @spond, Thank you very much for debugging the error. It works now. Since the problem is due to an internal HyPhy bug, do I need to do the analysis for all of my alignments or just those that I ran into the problem? Thanks, Alborz

spond commented 8 years ago

Dear @alborznoorani,

This error is an edge case which will always result in NaN results. If your other files come back with valid numeric values, then the issue does not apply.

Best, Sergei