Cytocast predicts variations in the human complexome


Most cellular processes are regulated by groups of proteins interacting together to form protein complexes. Protein compositions vary between different tissues or disease conditions enabling or preventing certain protein− protein interactions and resulting in variations of the complexome. Quantitative and qualitative characterization of context-specific protein complexes will help to better understand context-dependent variations in the physiological behavior of cells. Here we show how our whole cell simulation can link proteomics datasets with phenotype by identifying the key protein complexes associated to each phenotype.

This analysis is based on a few example simulations, where we take the proteome-wide protein abundances of 17 cell types from Kim et al. Each input dataset corresponds to a separate condition for Cytocast analysis.
We ran 2 stochastic simulations in Cytocast for all cell types and present here the results of these and the methodology to identify common patterns and key differences between cell types.

Explanation of menu points:

Single condition results / Protein complex prediction

Here one can find all the protein complexes our simulations predict for all the 17 cell types. We also report on these pages the abundances of each predicted complex together with the best matching reference complex from the CORUM database. This plot shows the number of predicted protein complexes for each condition (cell type).

All columns can be used to sort complexes. This could be done based on size of the simulated and the corresponding reference complexes, but also can rank them by the matching score. The structure of each complex can be explored under the list by selecting a given complex.

Multiple conditions comparison / Protein complex variation

In order to identify how protein complexes vary across conditions (between cell types), we merged similar complexes and ranked them by the variation (CV = STD/Mean) in their abundance and in their structural similarity (SS=average overlap score between complexes). The results can be explored in the Multi conditions comparison -> protein complex variation tab. By ordering complexes based on their CV (ascending) we can identify on the top of the list the the protein complexes, which are present in similar abundance in all cell types. These complexes are expected to play a vital role for the survival of the cell. Among those we identify Nuclear pore complexes (NPCs). When clicking on a row the tool lists the cell types the given complex was found in the simulations, together with its complex number, that can help users to search for a specific complex back in the Single condition results. At the same time the tool shows the members of the given complex and their interactions.
If sorted by minimal CV one can identify complexes that can be relevant to perform biological functions that are specific for a few or a single cell type. These cell type specific targets might be considered as potential targets for treatment. For instance, the voltage-gated potassium channel complex can be identified only in cells obtained from the adult frontal lobe tissues.
As expected the CD4+ T cells activation complex was identified only in Th cells
Cytocast doesn't stop here.
One can repeat the simulations, excluding selected target proteins to predict the effect of knocking out the given gene. Use cases on such scenario will be released soon. Thanks to such combined prediction of target and knockout effect, one will be able to identify and test a therapy before investing in tedious experiments.

Want to do more?

If you are interested to test Cytocast on your data, please contact us through the contact form of our website

Protein complex prediction

Export protein list

Protein prediction

Protein complex variation

Export protein list

Protein variation