AScorePro Error: "Input string was not in a correct format"

[lfu46]

Hi,

I'm encountering an error when attempting to localize modifications using AScorePro. The error message states: "An error occurred: Input string was not in a correct format."

I've attached the following for your review:

• Screenshot of the error
• options.json configuration file
• Peptide input file

Could you help me identify what's causing this error?

Thank you.

Best,
Longping

options.json

Eclipse_LF_OGlycoTM_HEK293T_OG_1_10062025 copy 2.tsv

[jmintser]

If the screenshot is showing the full command, then it seems you are missing some parameters. Please take a closer look at the README. The format should be
AScorePro.exe -j options.json -s scans.mzXML -p peptides.tsv
You might skip the -s option if you specify the mzXML in the options.json but it seems you do not

[lfu46]

Thank you for pointing that out. I have checked the README and confirmed that I explicitly specified the paths for both the raw scan files and peptide TSV files. However, the issue persists.

Could this be a format issue with my peptide TSV file?

Thank you for your help.

Best,
Longping

[rrad42]

Hello.

It wasnt obvious from the error but the main issue was that the ID column of the peptide was trying to be parsed as an integer. There's no reason it can be any arbitrary unique text, so the update now allows strings. There was also an issue with command line parameters vs paths specified on the command line. The latest will also follow the expected behavior where it will look in the options.json first for paths, but can be overridden by specifying CLI paths.

Please download the latest and give it a try. If there are still problems, it would help if you can share the raw file as well.

[lfu46]

Hi,

Thanks for your reply. The algorithm is now working, but I've identified an issue with peptide sequence modification. The algorithm is altering my input sequences, which is causing incorrect site assignments.
For example, spectrum Eclipse_LF_OGlycoTM_HEK293T_OG_1_10062025.13691.13691.3 shows:

• Input: W.TC$ST^SATTTAPSNSTPIAR.Y
• Output: W.TC^$STSATTTAPS^NSTPIAR.Y

The output sequence has relocated the modification markers, causing AScorePro to assign the site score to C at position 2 instead of the correct location. I've highlighted several additional examples in yellow in the attached figures.

Could you explain why the algorithm is modifying the input sequences? I've attached both the input and output files for reference.

out.tsv
Eclipse_LF_OGlycoTM_HEK293T_OG_1_10062025_528_copy.tsv
options.json

Thank you.

Best,
Longping

[rrad42]

Hi Longping,

This is part of the Ascore algorithm where the first step is to re-score the full peptide with the target mod on different positions. It will defer to the original position if no alternative scores better, but its not unusual for the mod to move around.

This is important to the site scoring because the difference in fragment ions between the first and second ranked peptides are used for the site localization score, and the first peptide should provide more matching fragment ions.

if the original site positions are needed, you can match them based on the ID column.

Ramin

[lfu46]

Hi Ramin,

Thank you for your clarification. Does this mean the original identification results are incorrect?

I used MSFragger for identification, which relies on b, y, c, and z ions. During identification, MSFragger generates a score for each possible modified peptide and reports the result with the highest score.

After AScorePro rescoring, the modification sites changed. This suggests that AScorePro and MSFragger give different identification and site localization results on the same spectrum. How should I determine which result is more reliable?

Additionally, is there a parameter that prevents AScorePro from relocalizing modification sites and instead generates a score only for the sites originally identified?

Thank you for your help.

Best,
Longping