search

MannLabs / directlfq

Fast and accurate label-free quantification for small and very large numbers of proteomes
https://doi.org/10.1101/2023.02.17.528962
Apache License 2.0
37 stars 4 forks source link

Minimum number for peptides for LFQ #13

Open pisistrato opened 1 year ago

pisistrato commented 1 year ago

Hi @ammarcsj ,

I am back with an additional question :) Is there a way to specify the minimum number of peptides a protein should have to get a LFQ value? Neither in the GUI nor in the code I don't see a filter on this, am I wrong? It would also be nice to get in the output table which are the peptides that were used for the LFQ calculation.

ammarcsj commented 1 year ago

Hi @pisistrato :)

indeed there is no filter for number of peptides, only for number of intensities with the following switch (I copy from the Github): -min_nonan: Min number of ion intensities necessary in order to derive a protein intensity. Increasing the number results in more reliable protein quantification at the cost of losing IDs.

So if you have DIA data and use all fragment intensities, you will have quite a few intensities already if you detect a peptide. If you use precursor or peptide intensities as input, you will have only as many as there are charge states, or even only one. However I found you can tune it quite easily for your dataset and its a bit more general than peptides (which would indeed be nice to have)

Does this already help?

Concerning the output table: indeed this would make sense. What you can only do currently, is look into the .ion_intensities.tsv table, where all the underlying data is stored. If a protein intensity is reported we currently always use all the data listed in the table.

Best Constantin