humap_complexes — hu.MAP human protein complexes, versions 1.0, 2.0 and 3.0 (2017-2024)

Description

Networks of human protein complexes and protein interactions from the three versions of hu.MAP, a map of human protein complexes obtained by integrating mass spectrometry proteomic experiments: hu.MAP 1.0 (2017; more than 9,000 experiments: co-fractionation, and affinity purification of tagged proteins of BioPlex and of Hein et al.), hu.MAP 2.0 (2020/2021; more than 15,000 experiments, adding proximity labeling (BioID), RNA pull-down and further affinity purification data) and hu.MAP 3.0 (2024; more than 25,000 experiments). In each version, a machine learning classifier, trained on known complexes (CORUM in 1.0 and 2.0, the Complex Portal in 3.0), gives the probability that two proteins interact (are in the same complex); the network of these interactions was then clustered, in two stages and with several parameters, to identify the complexes.

The networks are named by the version (v1_, v2_, v3_) and the kind of network:

v1_complexes, v2_complexes, v3_complexes: bipartite network of proteins and the complexes in which they are members (a protein can be in several complexes): 4,659 complexes in 1.0, 6,965 in 2.0 and 15,326 in 3.0, with the confidence of each complex in 2.0 and 3.0.
v1_complex_pairs, v3_complex_pairs: network of the pairs of proteins of the same complexes, weighted by the predicted probability that they interact. (hu.MAP 2.0 does not distribute this network as a file, only within its Cytoscape session; it is left out.)
v2_all_pairs, v3_all_pairs: network of all the pairs of proteins for which the classifier gave a prediction (of August 2020 and May 2023), weighted by the predicted probability that they interact; most pairs have a low probability. (hu.MAP 1.0 does not distribute these predictions, only the feature matrix.)

The training and test sets of complexes and interactions, the feature matrices, the enrichment tables and the Cytoscape sessions are left out. The data of hu.MAP 2.0 are kept as given, with their repetitions, which come from proteins (gene identifiers) mapped to the same UniProt accession: 34 complexes of v2_complexes list a protein twice (parallel edges), and in v2_all_pairs 66,458 pairs of proteins appear twice, with different probabilities, and 21 pairs join a protein to itself (self-loops).

The networks have the following properties:

(Node) uniprot: the UniProt accession of the protein (empty for complexes, and for the proteins of hu.MAP 1.0 without one; in 2.0 and 3.0, the identifier given instead of an accession for a few proteins, e.g. an NCBI Gene identifier).
(Node) gene_id (v1_*): the identifier of the protein in hu.MAP 1.0: its NCBI Gene identifier (an Ensembl gene identifier for 40 proteins).
(Node) name: the gene name of the protein (several names separated by ; in 3.0; empty if not given, e.g. for the proteins of the all_pairs networks that are in no complex), or the identifier of the complex (e.g. HuMAP2_00000, huMAP3_00000.1; in 1.0, the number of the complex, from 0, in the order of the file).
(Node) is_protein (*_complexes): whether the node is a protein (otherwise it is a complex).
(Node) confidence (v2_complexes, v3_complexes): the confidence of the complex: the number of the most stringent of the clusterings in which it was found, from 1 (extremely high) to 5 (medium) in 2.0, and to 6 (medium) in 3.0 (0 for proteins).
(Edge) probability (*_pairs): the probability that the two proteins interact, according to the classifier.

Tags

Biological Protein interactions Weighted Bipartite Multigraph Metadata

Added

2026-10-31

Citation

K. Drew, C. Lee, R. L. Huizar, F. Tu, B. Borgeson, C. D. McWhite, Y. Ma, J. B. Wallingford, E. M. Marcotte, "Integration of over 9,000 mass spectrometry experiments builds a global map of human protein complexes." Molecular Systems Biology 13, 932 (2017) 10.15252/msb.20167490 Sci-Hub
K. Drew, J. B. Wallingford, E. M. Marcotte, "hu.MAP 2.0: integration of over 15,000 proteomic experiments builds a global compendium of human multiprotein assemblies." Molecular Systems Biology 17, e10016 (2021) 10.15252/msb.202010016 Sci-Hub
S. N. Fischer, E. R. Claussen, S. Kourtis, S. Sdelci, S. Orchard, H. Hermjakob, G. Kustatscher, K. Drew, "hu.MAP3.0: atlas of human protein complexes by integration of >25,000 proteomic experiments." Molecular Systems Biology 21 (2025) 10.1038/s44320-025-00121-5 Sci-Hub

Upstream URL OK

https://humap3.proteincomplexes.org/
https://humap3.proteincomplexes.org/download
https://humap2.proteincomplexes.org/download
https://zenodo.org/records/15293715
https://hu1.proteincomplexes.org/download

Upstream license

Creative Commons Zero (CC0), credit requested (hu.MAP 2.0 files on Zenodo: CC BY 4.0), https://creativecommons.org/publicdomain/zero/1.0/

Networks

Tip: click on a column header to sort the list, or hover over it for a description.

Column legend

$\left<k\right>$: Mean degree
$\sigma_k$: Standard deviation of the degrees
$\lambda_h$: Largest eigenvalue of the Hashimoto (non-backtracking) matrix
$\tau$: Random walk mixing time
$r$: Degree assortativity
$c$: Global clustering coefficient
$\oslash$: (Pseudo-) diameter
$S$: Relative size of the largest component
NPs: Node properties
EPs: Edge properties

Name	Nodes	Edges	$\left<k\right>$	$\sigma_k$	$\lambda_h$	$\tau$	$r$	$c$	$\oslash$	$S$	Kind	Mode	NPs	EPs	gt	GraphML	GML	csv
v1_complexes	12,437	18,611	2.99	3.66	11.43	3759.76	-0.05	0.00	57	0.57	Undirected	Bipartite	is_protein uniprot gene_id name		291 KiB	419 KiB	415 KiB	355 KiB
v1_complex_pairs	7,778	64,048	16.47	37.71	183.94	1873.68	0.88	0.85	28	0.55	Undirected	Unipartite	uniprot gene_id name	probability	501 KiB	714 KiB	721 KiB	622 KiB
v2_complexes	16,928	30,802	3.64	3.33	12.10	423.07	-0.01	0.00	33	0.85	Undirected	Bipartite	is_protein confidence uniprot name		370 KiB	553 KiB	539 KiB	471 KiB
v2_all_pairs	15,373	17,526,311	2280.14	2484.03	4508.58	1.44	-0.38	0.54	5	1.00	Undirected	Unipartite	uniprot name	probability	72.7 MiB	119.8 MiB	118.6 MiB	96.9 MiB
v3_complexes	29,095	75,531	5.19	5.12	20.14	357.90	-0.02	0.00	30	0.97	Undirected	Bipartite	is_protein confidence uniprot name		616 KiB	1000 KiB	981 KiB	867 KiB
v3_complex_pairs	13,687	159,451	23.30	28.63	100.83	144.26	0.67	0.73	17	0.96	Undirected	Unipartite	uniprot name	probability	1.1 MiB	2.1 MiB	2.1 MiB	2.1 MiB
v3_all_pairs	17,233	25,992,007	3016.54	3126.21	5532.19	1.44	-0.45	0.51	5	1.00	Undirected	Unipartite	uniprot name	probability	111.2 MiB	216.6 MiB	213.7 MiB	226.1 MiB

Ridiculograms^*

v1_complex_pairs drawing — v1_complex_pairs

v3_complex_pairs drawing — v3_complex_pairs

^* These are automatically generated force-directed visualizations, and can be quite meaningless for networks both big and small. They should not be taken seriously as sources of scientific insight. See here for a discussion.