Beyond Expression Similarity: Contrastive Learning Recovers Functional Gene Associations from Protein Interaction Structure

• URL: http://arxiv.org/abs/2603.20955v1
• Date: Sat, 21 Mar 2026 21:36:23 GMT
• Title: Beyond Expression Similarity: Contrastive Learning Recovers Functional Gene Associations from Protein Interaction Structure
• Abstract summary: The Predictive Associative Memory (PAM) framework posits that useful relationships often co-occur in shared contexts.<n>We test whether this principle transfers to molecular biology, where protein-joint interactions provide functional associations.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: The Predictive Associative Memory (PAM) framework posits that useful relationships often connect items that co-occur in shared contexts rather than items that appear similar in embedding space. A contrastive MLP trained on co-occurrence annotations--Contrastive Association Learning (CAL)--has improved multi-hop passage retrieval and discovered narrative function at corpus scale in text. We test whether this principle transfers to molecular biology, where protein-protein interactions provide functional associations distinct from gene expression similarity. Four experiments across two biological domains map the operating envelope. On gene perturbation data (Replogle K562 CRISPRi, 2,285 genes), CAL trained on STRING protein interactions achieves cross-boundary AUC of 0.908 where expression similarity scores 0.518. A second gene dataset (DepMap, 17,725 genes) confirms the result after negative sampling correction, reaching cross-boundary AUC of 0.947. Two drug sensitivity experiments produce informative negatives that sharpen boundary conditions. Three cross-domain findings emerge: (1) inductive transfer succeeds in biology--a node-disjoint split with unseen genes yields AUC 0.826 (Delta +0.127)--where it fails in text (+/-0.10), suggesting physically grounded associations are more transferable than contingent co-occurrences; (2) CAL scores anti-correlate with interaction degree (Spearman r = -0.590), with gains concentrating on understudied genes with focused interaction profiles; (3) tighter association quality outperforms larger but noisier training sets, reversing the text pattern. Results are stable across training seeds (SD < 0.001) and cross-boundary threshold choices.

Related papers

• BioM-JEPA: joint-embedding prediction of graph-connected gene blocks in single cells [64.21963170814706]
  BioM-JEPA predicts aggregate representations of graph-connected gene blocks.<n>A student network infers each target-block representation from the remaining genes in a cell.<n>Across CellBench tasks, BioM-JEPA embeddings retained expression, pathway and neighbourhood information.
  arXiv  Detail & Related papers  (2026-08-06T11:58:29Z)
• Stable-Shift: Biologically Structured Prediction of Transcriptional Responses to Unseen Gene Perturbations [8.089217380596391]
  Stable-Shift is a structured method for estimating unseen-gene responses.<n>System combines STRING interactions, network structure, control-cell expression statistics, and Gene Ontology annotations.<n>Mean cosine similarity over five unseen-gene splits was 0.589 +/- 0.008.
  arXiv  Detail & Related papers  (2026-06-22T19:48:07Z)
• Event Embedding of Protein Networks : Compositional Learning of Biological Function [0.15229257192293197]
  We train 64-dimensional representations on random walks from the human STRING interactome.<n>We find that compositional structure substantially improves pathway coherence.<n>These results indicate that enforced compositionality specifically benefits relational and compositional reasoning tasks in biological networks.
  arXiv  Detail & Related papers  (2026-04-01T13:53:35Z)
• Beyond Independent Genes: Learning Module-Inductive Representations for Gene Perturbation Prediction [48.80217316452559]
  scBIG is a module-inductive prediction framework that explicitly models coordinated gene programs.<n> scBIG consistently outperforms state-of-the-art methods, particularly on unseen and perturbation settings.
  arXiv  Detail & Related papers  (2026-02-03T16:43:40Z)
• TRIDENT: A Trimodal Cascade Generative Framework for Drug and RNA-Conditioned Cellular Morphology Synthesis [56.9460577864211]
  TRIDENT is a cascade generative framework that synthesizes realistic cellular morphology by conditioning on both the perturbation and the corresponding gene expression profile.<n> TRIDENT significantly outperforms state-of-the-art approaches, achieving up to 7-fold improvement with strong generalization to unseen compounds.
  arXiv  Detail & Related papers  (2025-11-23T04:43:27Z)
• A U-Statistic-based random forest approach for genetic interaction study [5.418369213731247]
  We propose a U-Statistic-based random forest approach, referred to as Forest U-Test, for genetic association studies with quantitative traits.<n>Through simulation studies, we showed that the Forest U-Test outperformed existing methods.
  arXiv  Detail & Related papers  (2025-08-19T06:22:20Z)
• Hierarchical Multi-Label Contrastive Learning for Protein-Protein Interaction Prediction Across Organisms [2.399426243085768]
  We propose HIPPO, a hierarchical contrastive framework for protein-protein interaction prediction.<n>The proposed approach incorporates hierarchical contrastive loss functions that emulate the structured relationship among functional classes of proteins.<n> Experiments on benchmark datasets demonstrate that HIPPO achieves state-of-the-art performance, outperforming existing methods and showing robustness in low-data regimes.
  arXiv  Detail & Related papers  (2025-07-03T15:41:04Z)
• Unlasting: Unpaired Single-Cell Multi-Perturbation Estimation by Dual Conditional Diffusion Implicit Bridges [68.98973318553983]
  We propose a framework based on Dual Diffusion Implicit Bridges (DDIB) to learn the mapping between different data distributions.<n>We integrate gene regulatory network (GRN) information to propagate perturbation signals in a biologically meaningful way.<n>We also incorporate a masking mechanism to predict silent genes, improving the quality of generated profiles.
  arXiv  Detail & Related papers  (2025-06-26T09:05:38Z)
• Statistical Verification of Linear Classifiers [76.95660509846216]
  We propose a homogeneity test closely related to the concept of linear separability between two samples.<n>We focus on establishing upper bounds for the test's emphp-value when applied to two-dimensional samples.
  arXiv  Detail & Related papers  (2025-01-24T11:56:45Z)
• CSGDN: Contrastive Signed Graph Diffusion Network for Predicting Crop Gene-phenotype Associations [6.5678927417916455]
  We propose a Contrastive Signed Graph Diffusion Network, CSGDN, to learn robust node representations with fewer training samples to achieve higher link prediction accuracy. We conduct experiments to validate the performance of CSGDN on three crop datasets: Gossypium hirsutum, Brassica napus, and Triticum turgidum.
  arXiv  Detail & Related papers  (2024-10-10T01:01:10Z)
• FGBERT: Function-Driven Pre-trained Gene Language Model for Metagenomics [46.189419603576084]
  FGBERT is a novel metagenomic pre-trained model that employs a protein-based gene representation as a context-aware tokenizer.<n>It demonstrates superior performance on metagenomic datasets at four levels, spanning gene, functional, bacterial, and environmental levels.
  arXiv  Detail & Related papers  (2024-02-24T13:13:17Z)
• Improved K-mer Based Prediction of Protein-Protein Interactions With Chaos Game Representation, Deep Learning and Reduced Representation Bias [0.0]
  We present a method for extracting unique pairs from an interaction dataset, generating non-redundant paired data for unbiased machine learning. We develop a convolutional neural network model capable of learning and predicting interactions from Chaos Game Representations of proteins' coding genes.
  arXiv  Detail & Related papers  (2023-10-23T10:02:23Z)
• State-specific protein-ligand complex structure prediction with a multi-scale deep generative model [68.28309982199902]
  We present NeuralPLexer, a computational approach that can directly predict protein-ligand complex structures. Our study suggests that a data-driven approach can capture the structural cooperativity between proteins and small molecules, showing promise in accelerating the design of enzymes, drug molecules, and beyond.
  arXiv  Detail & Related papers  (2022-09-30T01:46:38Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.