Restarting the searches

[llmlrnr]

Hello,

Many thanks for the great software and the detailed project description!

I am using XiSearch on a linux cluster and run out of available computing time when performing very complex searches. Is there a way to restart a search that was cancelled at some point or the only way to make it work is to restrict the search space as much as possible?

[lutzfischer]

Not directly - the best I could suggest is to filter the peaklist to only contain spectra without result. xiSEACH even comes with a tool for that.

java -cp /path/to/xiSearch.jar rappsilber.gui.localapplication.ScanFilter

you could load the search result into the third tab (ScanFilter) and select "Exclude Selected". On the first tab you give your original mgf file and a path to a new one and press run.

[lutzfischer]

How many variable modifications are you searching?

[llmlrnr]

Thanks a lot for the suggestion! I had cysteine-carbamidomethylation and methionine-oxidation as two variable modifications but it definitely makes sense to set Cys-carbamidomethylation as a fixed modification!

I think the searches also take too long because I include the NonCovalentBound crosslinker along with my desired crosslinker, but it is the only way how the search can be performed, right?

[lutzfischer]

unless your crosslinker is reacting to cystein you probably set cysteine-carbamidomethylation as fixed - and even then other ways would be better - like having an additional crosslinker that is carbamidomethylation-lighter then the actuall crosslinker and links to cysteine-carbamidomethylation.

Long run time could also come from lag of memory. If you can try to give it more memory

[grandrea]

If i can suggest- put your modifications and crosslinker modifications as variable linear, rather than full variable, it will help memory quite a bit. You can drop the noncovalent, especially if you used in sourcd fragmentation, but it can lead to false positive identifications.

…

On Fri, 7 Jun 2024, 18:26 Luca, ***@***.***> wrote: Thanks a lot for the suggestion! I had cysteine-carbamidomethylation and methionine-oxidation as two variable modifications but it definitely makes sense to set Cys-carbamidomethylation as a fixed modification! I think the searches also take too long because I include the NonCovalentBound crosslinker along with my desired crosslinker, but it is the only way how the search can be performed, right? — Reply to this email directly, view it on GitHub <#105 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AKR4YAYA73LPJC7UFDDT2MLZGHNKXAVCNFSM6AAAAABI64PZX2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNJVGE2TSOJWHE> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[llmlrnr]

Hello again,

Is there a way to perform peaklist splitting before starting the search?
So instead of filtering the peaklist after initial searches I would like to create 2-3 peakfiles out of one, use them for the search in parallel, and combine the results afterwards. Are you aware of any tools that can do that?

[grandrea]

Hi,
Yes you can use msconvert for this (subset by time or scan event), or simply write a script using pyteomics pyhton package to split the mgf files (chatgpt4 knows the pyteomics syntax quite well). You can then use the scripts in the hpc_scripts in this repo to run in parallel on a cluster (or whatever script you like of course!) and then combine_searches.py or the like to recombine the results.

[llmlrnr]

I guess this is exactly what I needed, thank you!!