14/14 And what about your protein? You can check the predictions, dynamics and moonlighting behaviour on our interactive platform! 😊 biosam.shinyapps.io/plasmapx_shi...

13/14 Our predictions identified 13 known moonlighters (such as MCM3/5) and predicted 25 new ones (like signal recognition particle subunit SRP9).

12/14 We also used our data to predict moonlighting subunits of protein complexes. Moonlighting subunits are known stable subunits of complexes that have an additional function outside of those complexes.

11/14 We found that most subunits are more likely to participate in their complex in the later stages of the blood cycle (with some notable exceptions!). The complex assemblies typically peak jusft after the peak of the abundances of the complex subunits.

10/14 By the way, we collected the mass spec data in biological triplicates and in 7 timepoints across the IDC, so we could also look at the dynamics of the interactions!

9/14 Thanks to our great collaborators from Tim Gilgeber’s lab @cssbhamburg.bsky.social and @nishaphilip.bsky.social @edinburgh-uni.bsky.social , we validated some more complexes using fluorescence microscopy and pull-downs, like an exported complex in Maurer’s clefts and MON1/CCZ1/RMC1 complex.

8/14 We also identified two missing subunits of RNA exosome ring! Their sequence has a low similarity to the orthologs in model eukaryotic organisms, but are structurally conserved based on AlphaFold 3 prediction.

7/14 But now to the results: We could reconstruct complexes that have already been previously studied, but also some that are conserved in eukaryotes and have not been dissected in Plasmodium yet, such as LSm spliceosome subcomplex.

6/14 (Yes, we made the MAPX pipeline available as an R package) github.com/SamPazicky/M...

5/14 The answer was: machine learning! We developed MAP-X (meltome-assisted profiling of protein complexes) that derives few parameters from melting curves and raw data and evaluates a protein-protein interaction for each pair of proteins. The pairwise data can be clustered to predict the complexes!

4/14 The question was, how to use the TPP data alone to predict new complexes?

3/14 Subunits that form stable complex precipitate in a similar manner – this was already known and used to track the dynamics of protein complexes. Below, each point represent a protein in the dataset and the points with the same colors are subunits of protein complexes - they precipitate together!

2/14 To do this, we used thermal proteome profiling (TPP) coupled with mass spec. This technique is not only simpler+cheaper than other interactome mapping methods, but also maps the interactions in live cells! TPP measures how the proteome precipitates in cells exposed to a temperature gradient.

And what about your protein? You can check the predictions, dynamics and moonlighting behaviour of your favourite protein on our interactive platform! 😊 biosam.shinyapps.io/plasmapx_shi...

Our predictions identified 13 known moonlighters (such as MCM3/5 subcomplex) and predicted 25 new ones (like signal recognition particle subunit SRP9).

You could do on-bead digestion to exclude the SDS. For bioID I used to do on-bead digestion and then 1st “elution” with AmBic and 2nd with 4:1 ACN:TFA. As for removal of SDS, I would be worried to go for anything less than SP3. Still weird that it’s only a particular treatment though!

Thank you, yes I was reading this paper just few days back. Our experience with FAIMS/DDA is already quite different than theirs (no boost in protein IDs but more unique peptides with FAIMS in my hands), so I assume FAIMS/DIA might also end up being different...

Thank you, this is what I was thinking too. Definitely will try library-free as well and stick with what works better.