A few weeks ago, I had the absolute pleasure of catching up with Jacob Schreiber on his podcast, The Bioinformatics Chat. Jacob and I met as grad students in Bill Noble’s lab at UW. Chronologically, we were in the same cohort (started grad school in 2014) but he was in the Computer Science and Engineering program and I was in the Genome Sciences program.
From day one, he absolutely blew me away with his knowledge and as a baby bioinformatician, I was so intimidated! For a few years, I barely understood what, exactly, he did. All I knew was that you should always add MORE LAYERS to your neural net. It turned our he, too, barely understood what, exactly, I did. Imposter syndrome is a hell of a drug.
I kicked off the scientific presentations for the virtualized Northeastern University May Institute with a workshop on the basics of mass spectrometry proteomics. In this mixed-methods 1.5 hour session, I aimed to give biomedical researchers a crash course in all things quantitative mass spectrometry-based proteomics and even give some of the pros a few tips on the Skyline software ecosystem. By the end of this workshop, I wanted participants to come away with the ability to:
Assess the experimental pros and cons of targeted proteomics, and compare to discovery proteomics.
Explain the fundamentals of mass spectrometry proteomics, including peptide fragmentation and basic components of a mass spectrometer
Describe the steps of a targeted proteomics workflow and the information required to build an assay
Apply the Skyline software ecosystem to their own targeted proteomics experiments.
Unsurprisingly, my most common approach to proteome abundance measurements by mass spectrometry is data independent acquisition (DIA). Specifically I’ve been using the chromatogram library approach (Searle et al 2018) because, compared to spectral library-based approaches, it doesn’t take a lot of extra work. I just prepare my samples as usual, then pool a few uL of each sample into a “library” or consensus sample. I queue up my single-shot experimental samples, then I acquire the pooled library sample with multiple injections, each time spanning a 100 m/z range (gas phase fractionation, GPF) with very narrow isolation windows.
The next step up is to search the narrow window, GPF multi-injections against a spectral library. Recently, a team of researchers released “Prosit”, a tool to predict spectral libraries. Using Prosit predicted spectral libraries to search GPF chromatogram libraries gives detection numbers a boost (Searle et al 2020). Because it’s so easy to use predicted spectral libraries, I’ve been doing it for all my projects.
The tutorial above is a work in progress, so let me know if you have questions or suggestions to improve it!
Learning to code, coming from an experimental background, can be a frustrating and intimidating experience. Although there are manyfreecoursesonline that will teach you the basics of Python and programming, the best way to improve is to simply practice. The best practice is a project you are personally motivated and passionate about. For graduate students, this might be a component of your thesis or a side project that complements your research interests. Coming up with an interesting project might be daunting at first, however, so here are some resources for quick, achievable practice problems to help keep you coding.
Students, teachers, parents — welcome! Thank you for your interest in following up on the Expanding Your Horizon’s workshop! Here’s a summary of what we talked about today, along with many links to videos and sites where you can learn more about gene editing.