Mutual Interactors as a Principle for Phenotype Discovery in Molecular Interaction Networks

Biological networks are powerful representations for the discovery of molecular phenotypes. Fundamental to network analysis is the principle—rooted in social networks—that nodes that interact in the network tend to have similar properties. While this long-standing principle underlies powerful methods in biology that associate molecules with phenotypes on the basis of network proximity, interacting molecules are not necessarily similar, and molecules with similar properties do not necessarily interact.

Here, we show that molecules are more likely to have similar phenotypes, not if they directly interact in a molecular network, but if they interact with the same molecules. We call this the mutual interactor principle and show that it holds for several kinds of molecular networks, including protein-protein interaction, genetic interaction, and signaling networks.

We develop a machine learning framework for predicting molecular phenotypes on the basis of mutual interactors. Strikingly, the framework can predict drug targets, disease proteins, and protein functions in different species, and it performs better than much more complex algorithms. The framework is robust to incomplete biological data and is capable of generalizing to phenotypes it has not seen during training. Our work represents a network-based predictive platform for phenotypic characterization of biological molecules.

Molecules in and across living cells are constantly interacting, giving rise to complex biological networks. These networks serve as a powerful resource for the study of human disease, molecular function and drug-target interactions. For instance, evidence from multiple sources suggests that causative genes from the same or similar diseases tend to reside in the same neighborhood of protein-protein interaction networks. Similarly, proteins associated with the same molecular functions form highly-connected modules within protein-protein interaction networks.

These observations have motivated the development of bioinformatics methods that use molecular networks to infer associations between proteins and molecular phenotypes, including diseases, molecular functions, and drug targets. Many of these methods assume that molecular networks obey the organizing principle of homophily: the idea that similarity breeds connection. However, while this principle has been well-documented in social networks of many types (e.g. friendship, work, co-membership), it is unclear whether it captures the dynamics of biological networks. If not, existing bioinformatics methods that assume homophily may not realize the full potential of biological networks for scientific discovery.

Motivation for Mutual Interactors

To better understand the place for homophily in bioinformatics, we consider groups of phenotypically similar molecules (e.g. molecules associated with the same disease, involved in the same function, or targeted by the same drug) and study their interactions in large-scale biological networks. We find that most molecules associated with similar phenotypes do not interact directly in molecular networks, a result which puts into question the assumption of homophily, an assumption that is taken for granted by so many bioinformatics methods.

In fact, a different principle better explains how phenotypic similarity relates to network structure in biology. On average, two molecules that interact directly with one another will have less in common than two molecules that share many mutual interactors, just as people in a social network may share mutual friends. We call this the mutual interactor principle and validate it empirically on a diverse set of biological networks (see Figure 1c).

Mutual Interactors Approach

Motivated by our findings, we develop a machine learning framework, Mutual Interactors, that can predict a molecule’s phenotype based on the mutual interactors it shares with other molecules. We demonstrate the power, robustness, and scalability of Mutual Interactors on three key prediction tasks: disease protein prediction, drug target identification, and protein function prediction. With experiments across three different kinds of molecular networks (protein-protein interaction, signaling and genetic interaction) and four species (H. sapiens, S. cerevisiae, A. thaliana, M. musculus), we find that Mutual Interactors substantially outperforms existing methods, with gains in recall up to 61%. Additionally, we show that the weights learned by our method provide insight into the functional properties and druggability of mutual interactors.

Mutual Interactors is an approach based on a different network principle than existing bioinformatics methods. That it can outperform state-of-the-art approaches suggests a need to rethink the fundamental assumptions underlying machine learning methods for network biology

Uncovering Disease-Driving Proteins Using Mutual Interactors

Identifying Drug Targets Using Mutual Interactors

Publication

Mutual Interactors as a Principle for Phenotype Discovery in Molecular Interaction Networks
Sabri Eyuboglu*, Marinka Zitnik*, and Jure Leskovec
Pacific Symposium on Biocomputing, PSB 2023

@inproceedings{eyuboglu2023mutual,
title = {Mutual Interactors as a Principle for Phenotype Discovery in Molecular Interaction Networks},
author = {Eyuboglu, Sabri and Zitnik, Marinka and Leskovec, Jure},
booktitle = {Pacific Symposium on Biocomputing, PSB},
year      = {2023}
}

Code

Python implementation of Mutual Interactors are available in the GitHub repository.

Authors

Latest News

Mar 2025:   On Biomedical AI in Harvard Gazette

Read about AI in medicine in the latest Harvard Gazette and New York Times.

Mar 2025:   TxAgent: AI Agent for Therapeutic Reasoning

TxAgent is an AI agent for therapeutic reasoning that consolidates 211 tools from trusted sources, including all US FDA-approved drugs since 1939 and validated clinical insights. [Project website] [TxAgent] [ToolUniverse]

Mar 2025:   Multimodal AI predicts clinical outcomes of drug combinations from preclinical data

Mar 2025:   KGARevion: AI Agent for Knowledge-Intensive Biomedical QA

KGARevion is an AI agent designed for complex biomedical QA that integrates the non-codified knowledge of LLMs with the structured, codified knowledge found in knowledge graphs. [ICLR 2025 publication]

Feb 2025:   MedTok: Unlocking Medical Codes for GenAI

Meet MedTok, a multimodal medical code tokenizer that transforms how AI understands structured medical data. By integrating textual descriptions and relational contexts, MedTok enhances tokenization for transformer-based models—powering everything from EHR foundation models to medical QA. [Project website]

Feb 2025:   What If You Could Rewrite Biology? Meet CLEF

What if we could anticipate molecular and medical changes before they happen? Introducing CLEF, an approach for counterfactual generation in biological and medical sequence models. [Project website]

Feb 2025:   Digital Twins as Global Health and Disease Models

Jan 2025:   LLM and KG+LLM agent papers at ICLR

Jan 2025:   Artificial Intelligence in Medicine 2

Excited to share our new graduate course on Artificial Intelligence in Medicine 2.

Jan 2025:   ProCyon AI Highlighted by Kempner

Thanks to Kempner Institute for highlighting our latest research, ProCyon, our protein-text foundation model for modeling protein functions.

Jan 2025:   AI Design of Proteins for Therapeutics

Dec 2024:   Unified Clinical Vocabulary Embeddings

New paper: A unified resource provides a new representation of clinical knowledge by unifying medical vocabularies. (1) Phenotype risk score analysis across 4.57 million patients, (2) Inter-institutional clinician panels evaluate alignment with clinical knowledge across 90 diseases and 3,000 clinical codes.

Dec 2024:   SPECTRA in Nature Machine Intelligence

Are biomedical AI models truly as smart as they seem? SPECTRA is a framework that evaluates models by considering the full spectrum of cross-split overlap: train-test similarity. SPECTRA reveals gaps in benchmarks for molecular sequence data across 19 models, including LLMs, GNNs, diffusion models, and conv nets.

Nov 2024:   Ayush Noori Selected as a Rhodes Scholar

Congratulations to Ayush Noori on being named a Rhodes Scholar! Such an incredible achievement!

Nov 2024:   PocketGen in Nature Machine Intelligence

Oct 2024:   Activity Cliffs in Molecular Properties

Oct 2024:   Knowledge Graph Agent for Medical Reasoning

Sep 2024:   Three Papers Accepted to NeurIPS

Exciting projects include a unified multi-task time series model, a flow-matching approach for generating protein pockets using geometric priors, and a tokenization method that produces invariant molecular representations for integration into large language models.

Zitnik Lab  ·  Artificial Intelligence in Medicine and Science  ·  Harvard  ·  Department of Biomedical Informatics