Though I got my start in proteomics and bioinformatics late in my Ph.D. (c.2007), my post-doc is where I really got into coding. At the time this was a lot of handmade NNs in Matlab (re-writing Jonas Almeida’s code was my first job), and later more SVMs, RFs, kNN, etc., all back in early 2010s. So calling 2022 “The Year of ML in Proteomics” is specific to my own efforts coming to a head in 2022, not that the field has arrived (it was chugging along before and continues to chug along today), but this makes for a nice 2022 snapshot of ML with respect to proteomics applications.
All of this started back in early 2021 with a proposal Magnus Palmblad asked me to work on with him (he did most of it), which we later turned into a Lorentz Center Workshop with two other organizers, Viktoria Dorfer and Lennart Martens. This workshop was spectacular and combined experts from all over (and me). You can find some pics on twitter using the #ProteomicsML hashtag. Check the author list on the white paper below to see who was there… yeah, it was AWESOME!
More amazing, the workshop proposed ideas and actually did them!
Not to gush, but I am so happy with all of these outputs (all with Magnus Palmblad):
- ProteomicsML.org (and paper)
- 2022 ACS Measurement Symposium Session “Deep Learning in Proteomics: Deep Insight or Deeper Pitfalls?”
- JPR Virtual Issue on ML in Proteomics and Metabolomics
- THE white paper capturing the concepts discussed and future visions from the workshop. “We can either save the world or become evil billionaires.” - participant from workshop on last day
I hope there are a couple more papers/projects coming out (I have one I really want to see which back-calculates the LC gradient agnostically), but overall it was incredible to get to be around all of these scientists, and also now having new friends that are way smarter than me who I can call up and not be completely awkward around.