Wow. Okay, of course this is one way to do it! Actually, it seems blindingly obvious to me right this second, but I certainly hadn't thought to do it the way they did in this new study!
There are several smart ways to get around proteogenomic challenges that we're increasingly being encouraged to think about. It's a lot easier if we can just use the UniProt SwissProt little tiny library of protein sequences, but they ignore:
mutations,
individual biological variation,
the fact that some of those sequences might have been Craig Venter's and if you were picking a normal human being to pull sequences from, would he really be the best pick?
What this group does? They pull out their little nanopore gizmos and they do some sequencing. Instead of feeding their Variant Call File (vcf) or whatever is the same thing in Nanopore world into a search engine, they feed those sequences to deep learning peptide fragment and creation tools.
BOOM! Modeled MS/MS spectra that they can match against theoretical! Smart, right?
No comments:
Post a Comment