High-throughput immunopeptidomics

In my PhD I focused on studying the complexity of the immune system at the level of the T cell repeptor. Recently I’ve been getting in to what happens on the other side of the conversation as well; in addition to looking at TCR repertoires I’m increasingly playing with MHC-bound peptide repertoires too.

Immunopeptidomics is a super interesting field, with a great deal of promise, but it’s got a much higher barrier to entry for research groups relative to something like AIRR-seq. Nearly every lab can do PCR, and access to deep-sequencing machines or cores becomes ever cheaper and more commonplace. However not every lab has expertise with fiddly pull downs, while only a tiny fraction can do highly sensitive mass spec. This is why efforts to make immunopeptide data generation and sharing easier should be suitably welcomed.

One of the groups whose work commendably contributes to both of these efforts is that of Michal Bassani-Sternberg. For sharing, she consistently makes all of her data available (and is seemingly a senior founder and major contributor to the recent SysteMHC Atlas Project), while for generation her papers give clear and thorough technical notes, which aid in reproducibility.

However from the generation perspective this paper (which came out at the end of last year in Mol. Cell Proteomics) describes a protocol which – through application of sensible experimental design – should result in the easier production of immunopeptidomic data, even from more limited samples.

The idea is to basically increase the throughput of the methods by hugely reducing the number of handling steps and time required to do the protocol. Samples are mushed up, lysed, spun, and then run through a variety of stacked plates. The first (if required) catches irrelevant, endogenous antibodies in the lysates; the next catches MHC class I (MHC-I) peptide complexes via bead-cross-linked antibodies; the next similarly catches pMHC-II, while the final well catches everything else (giving you lovely sample-matched gDNA and proteomes to play with, should you choose). Each plate of pMHC can then be taken and treated with acid to elute the peptides from their grooves, before purification and mass spec. It’s a nice neat solution, which supposedly can all be done with readily commercially available goodies (although how much all these bits and bobs cost I have no idea).

Crucially it means that you get everything you might want (peptides from MHC-I/-II, plus the rest of the lysates) in separate fractions, from a single input sample, in a protocol that spans hours rather then days. Having it all done in one pass helps boost recovery from limited samples, which is always nice for say clinical material. Although I should say, ‘limited’ is a relative term. For people used to dealing with nice, conveniently amplifiable nucleic acids, tens to thousands of cells may be limiting. Here, they managed to go down as low as 10 million. (Which is not to knock it, as this is still much much better then hundreds of millions to billions of cells which these experiments can sometimes require. I don’t want everyone to go away thinking about repurposing their collection of banked Super Rare But Sadly Impractically Tiny tissue samples here.)

So on technical merit alone, it’s already a pretty interesting paper. However, there’s also a nice angle where they test out their new protocol on an ovarian carcinoma cell line with or without IFNg treatment, which tacks on a nice bit of biology to the paper too.

You see the things you might expect – like a shift in peptides seemingly produced by degradation from the standard proteasome to more of those produced by the immunoproteasome – and some you might not. Another nice little observation which follows on perfectly from this is that you also see an alteration in the abundance of peptides presented by different HLA alleles: for instance the increased  chemotryptic-like degradation of the immunoproteasome favours the loading of HLA-B*07:02 molecules, due to making more peptides with the appropriate motif.

My favourite observation however relates to the fact that there’s a consistent quantitative and qualitative shift in peptidomes between IFNg treated cells and mock. This raises an interesting possibility to me, about what should be possible in the near future, as we iron out the remaining wrinkles in the methodologies. Not only should we learn about what proteins are being expressed, based on which proteins those peptides are derived from, but we should be able to infer something about what cytokines those cells have been expressed to, based on how those peptides have been processed and presented.

Bulk downloading proteome files from UniProt using Python

It's that time again, where the following has happened:

1. I want to do some niche bioinformatics related thing
2. I cobble together a quick script to do said thing
3. I throw that script up on the internet on the offchance it will save someone else the time of doing 2

It's a little shift of target and scale from a similar previous post (in which I used Python to extract specific DNA sequences from UCSC. This time I've been downloading a large number of proteome files from UniProt.

It's all explained in the docstring, but the basic idea is that you go on UniProt, search for the proteomes you want, and use their export tool to download tsv files containing the unique accession numbers with identify the data you're after. Then you simply run this script in the same directory; it takes those accessions, turns them in to URLs, downloads the FASTA data at that address and outputs it to new FASTA files on your computer, with separate files named after whatever the tsv files were named.

The best thing about this is you can download multiple different lists of accessions, and have them output to separate files. Say maybe you have a range of pathogens you're interesting in, each with multiple proteomes banked; this way you end up with one FASTA file for each, containing as many of their proteomes as you felt like including in your search.