Truace
Subscribe

Interpretable neural networks prioritize cancer driver genes from genome-wide dependency landscapes

Identifying cancer driver genes and their therapeutic impact remains a core challenge in computational cancer biology. We introduce xNNDriver and xAEDriver, two interpretable neural network frameworks that connect cancer mutations with genome-wide DepMap gene dependencies, pathway activity, and drug-response patterns. xNNDriver is a supervised pathway-guided model that evaluates whether a gene's mutation status is encoded in the genome-wide dependency landscape; we interpret model fitness as a driver potential s…

By the Truvace desk2 min read1 sourcePeer-reviewedTRV-2026-2038· cite & verify
Interpretable neural networks prioritize cancer driver genes from genome-wide dependency landscapes
Polymerized-Laminin-332-Matrix-Supports-Rapid-and-Tight-Adhesion-of-Keratinocytes-Suppressing-Cell-pone.0035546.s008 by Kariya Y, Sato H, Katou N, Kariya Y, Miyazaki K. CC BY 2.5 · https://creativecommons.org/licenses/by/2.5

In brief

Researchers built two interpretable deep learning frameworks, xNNDriver and xAEDriver, that connect cancer mutations to genome-wide DepMap dependency data, pathway activity and drug-response patterns. The supervised model scores how strongly a gene's mutation status is reflected in the dependency landscape, while the unsupervised autoencoder learns latent binary features called Driver Variant Representations.

The work matters because it offers a way to prioritize candidate drivers, pathways and therapeutic vulnerabilities in computational cancer biology. The authors present it only as hypothesis-generating, so the ranked genes and associations still depend on later experimental validation before they could inform treatment.

Main points

  1. xNNDriver is a supervised, pathway-guided model whose fit is read as a driver potential score for each gene.
  2. It was applied to 3,008 candidate genes and recovered major established drivers alongside literature-supported candidates.
  3. Pathway analyses pointed to programs involving metabolism, growth factor signaling and immune regulation.
  4. xAEDriver's Driver Variant Representations captured cell-line-specific dependency and expression patterns linked to drug sensitivity.

The gain

Two interpretable neural network models linked mutations to gene dependencies and recovered established cancer drivers while flagging literature-supported candidates across 3,008 genes.

The rundown

The first model, xNNDriver, tests whether a gene's mutation status is encoded in the genome-wide dependency landscape. Its fitness is interpreted as a driver potential score that quantifies the mutation-dependency signal and favors genes with broad functional footprints.

The second model, xAEDriver, is an unsupervised autoencoder guided by the frequency distribution of known driver mutations. It learns latent binary features, called DVRs, that reflect combinatorial functional states in cell lines.

The authors report that DVRs are associated with drug sensitivity and pathway activity. They conclude that dependency landscapes carry interpretable signals of oncogenic function, offering a framework for prioritizing targets for further study.

Sources

  1. Peer-reviewedPLOS Computational Biology2026-10-13

The debate