Producing predicted library from a tsv file

[yangkl96]

Hi Vadim,

I am trying to predict a library with a given input tsv to produce a predicted.bin with the newest DIA-NN. I get the following error on Windows and Linux machines as of version 1.9.1. Versions 1.9 and earlier work fine. How can I fix this?

2024-10-30 10:51:59 [INFO] - diann-linux --lib spectraRT.tsv --predict --threads 11 --strip-unknown-mods --predict-n-frag 100
2024-10-30 10:52:00 [INFO] - DIA-NN 1.9.2 (Data-Independent Acquisition by Neural Networks)
2024-10-30 10:52:00 [INFO] - Compiled on Oct 20 2024 02:59:53
2024-10-30 10:52:00 [INFO] - Current date and time: Wed Oct 30 10:52:00 2024
2024-10-30 10:52:00 [INFO] - Logical CPU cores: 32
2024-10-30 10:52:00 [INFO] - Predicted spectra will be saved in a binary format
2024-10-30 10:52:00 [INFO] - Thread number set to 11
2024-10-30 10:52:00 [INFO] - WARNING: unrecognised option [--strip-unknown-mods]
2024-10-30 10:52:00 [INFO] - Deep learning predictor will predict 100 fragments
2024-10-30 10:52:00 [INFO] - WARNING: protein inference is enabled but no FASTA provided - is this intended?
2024-10-30 10:52:00 [INFO] -
2024-10-30 10:52:00 [INFO] - 0 files will be processed
2024-10-30 10:52:00 [INFO] - [0:00] Loading spectral library spectraRT.tsv
2024-10-30 10:52:00 [INFO] - [0:00] Finding proteotypic peptides (assuming that the list of UniProt ids provided for each peptide is complete)
2024-10-30 10:52:00 [INFO] - [0:00] Spectral library loaded: 0 protein isoforms, 0 protein groups and 20771 precursors in 18425 elution groups.
2024-10-30 10:52:00 [INFO] - ERROR: library contains fragments with invalid/unspecified charges

My spectraRT.tsv looks like this:
peptide charge
SPAAAAALIFPG 3

Thanks,
Kevin

[vdemichev]

Hi Kevin,

Thanks for letting me know, I will look into this. What happens is that in the --predict mode (special for FragPipe) DIA-NN just loads a precursor list and not the whole library, and then sometime after that a check for inappropriate fragment info kicks in. Will fix.

Best,
Vadim