Sequence-mask-search-hybrid thing finds more HLA peptide IDs!

I don't get the maths in this new study, but I do like getting more peptide identifications.

What is a sequence-mask-search hybrid de novo peptide sequencing framework? No idea. I assumed the 3 hyphenated words were a stats or math thing I've never heard of, but if it is, Google also hasn't heard of it.

I looked at all the Greek letters and decided to just download the supplemental material. They appear to train this Mask Search thing on some proteomics data and then they go after some HLA peptide data and find more peptides than anything else they search it with.

Exciting! Maybe this is the search engine and FDR for +1 peptides?!? At long last!

Unfortunately, while this group reports more identifications, the data they use isn't what I was hoping for. Out of 157,000 MS/MS scans searched, only 3,023 +1 ions were fragmented. This is a cool dataset, if all the peptides were +2 charged, peptidomics is WAY less scary!

What about the K/R found in this set?
Out of around 92,000 IDs

44,700 lysines
32,700 arginines

The average charge state of +2 checks out.

I'm not going to hit the big orange Publish button unless I can find something cooler in this paper -- back to the nerdy maths -- how did they support that all this work they did was worth publishing?

...okay...that's not too shabby, right? They went back to mass accuracy shifts at multiple positions as  a metric of match quality....I can get behind that enough to hit the button.

It's all Python stuff that can be downloaded here.