hu.MAP3.0: atlas of human protein complexes by integration of >25,000 proteomic experiments

What the paper is about

The paper introduces hu.MAP3.0, a giant “atlas” (map) of how human proteins come together into complexes. Proteins rarely act alone. Most of the work inside cells, such as making energy, repairing DNA, or responding to signals, occurs when proteins assemble into complexes, much like machines built from many parts. The challenge: We didn’t know the full “parts list” of these machines, or how most human proteins fit into them.

So, the researchers:

• Integrated over 25,000 mass spectrometry experiments (these detect which proteins physically stick together to form complexes).
• Used machine learning to stitch this massive data into a coherent map.
• Identified 15,000+ human protein complexes involving ~70% of all human proteins.
• Cross-checked these complexes with structural models from AlphaFold (AI-predicted 3D protein structures).

Why this is important

1. Completeness: This is the most comprehensive map to date of human protein complexes. 
2. Uncharacterized proteins: They gave functional clues for 472 proteins that were previously “mystery proteins.” Now we know who their partners are and can infer their roles.
3. Complex flexibility: They found ~6,000 cases where proteins are mutually exclusive (they compete for the same binding site). This means that complexes can change their roles depending on the context (i.e., different tissues, diseases, or conditions).
4. Public resource: The data is open for scientists through the hu.MAP3.0 website and EMBL-EBI’s Complex Portal, so anyone can use it.
   

How it helps in real life

This atlas has direct biomedical implications:

• Disease research: Many diseases (cancer, neurological disorders, rare genetic syndromes) arise when protein complexes break down. Now, scientists can pinpoint which subunit or interaction is likely involved.
• Drug discovery: Instead of targeting a single protein blindly, researchers can see the whole machine it belongs to and design drugs that disrupt or stabilize the right interactions.
• Unknown proteins: Those 472 newly placed proteins can now be investigated as potential disease genes or new drug targets.
• Precision medicine: Since complexes can switch subunits, the atlas explains why the same gene mutation might cause different outcomes in different tissues – information that can be critical for tailoring therapies.

Abstract

Macromolecular protein complexes carry out most cellular functions. Unfortunately, we lack the subunit composition for many human protein complexes. To address this gap we integrated >25,000 mass spectrometry experiments using a machine learning approach to identify >15,000 human protein complexes. We show our map of protein complexes is highly accurate and more comprehensive than previous maps, placing nearly 70% of human proteins into their physical contexts. We globally characterize our complexes using mass spectrometry based protein covariation data (ProteomeHD.2) and identify covarying complexes suggesting common functional associations. hu.MAP3.0 generates testable functional hypotheses for 472 uncharacterized proteins which we support using AlphaFold modeling. Additionally, we use AlphaFold modeling to identify 5871 mutually exclusive proteins in hu.MAP3.0 complexes suggesting complexes serve different functional roles depending on their subunit composition. We identify expression as the primary way cells and organisms relieve the conflict of mutually exclusive subunits. Finally, we import our complexes to EMBL-EBI’s Complex Portal and provide complexes through our hu.MAP3.0 web interface. We expect our resource to be highly impactful to the broader research community.