When searching files with a very small spectral libraries, (e.g. around 11000 precursors), some searches return almost empty or entirely empty pg.matrix.tsv files. The search does not fail and there are no errors, suggesting that this is a matter of how many precursor identifications pass the FDR threshold. This behavior can be reproduced by running fully predicted searches on an example file with increasingly small .fasta files. The issue is relevant for processing samples which have naturally few protein IDs, e.g. blood plasma or other body fluid samples.