Closed alessandro-vai closed 1 year ago
Hi @alessandro-vai, thank you for reporting. The point at which the error is happening suggests that there is an anomaly in the frequency distributions for that mass shift, most probably resulting in an abnormal value in error estimation. It's hard to guess what it is, and I can imagine that sharing all input data may not be very convenient, so I added a debug output at the right place which could help.
If you can update to the latest commit and run AA_stat again with -v 2
added to the command line, the debug lines before the error (let's say for -4.0308 Da and -2.0722 Da figures) would be really helpful.
Thanks for the prompt reply. Here it is the log. log_AA_stat.txt
Thank you! It looks like the error is happening when all of the error estimates for amino acid frequencies are undefined. When only some are, it's fine. I made a change to replace all NaN values with zero, which should not matter for the other cases. Can you please check the latest master?
Also, I notice that you seem to have very few decoys in the input. Is it filtered? AA_stat is designed to work with unfiltered data, so if possible, you can try that and see if the results improve.
It works! I guess the fact histones yield to small datasets causes troubles to the some fitting procedures. I noticed the very low number of decoys, but probably it is just right. I will double check the input files. Cheers
Hi,
I am running AA_stat on open search results of histone dataset. For this reason, I have fewer PSMs than I would get with a total proteome experiment, so the warning is perfectly fine. However, the tool fails at the "Plotting mass shift figures" step, producing the png plot files up to the mass shift -4.0308. Don't know if you need input files for the debugging.
Thank you in advance.