Therapeutic targets in lung diseases identified through single-cell analysis and mendelian randomization

Abstract

Background

The global rise in lung diseases necessitates identification of therapeutic targets, yet systematic studies integrating diverse pathologies and lung function traits to uncover causal targets remain limited.

Objectives

To identify causal therapeutic targets across 12 lung traits (10 diseases, 2 functional measures: Forced Vital Capacity [FVC] and the FEV1/FVC ratio) and validate their genetic roles.

Methods

Single-cell transcriptomic data from the Human Lung Cell Atlas were analyzed via Geneformer, a deep learning model, to simulate in silico perturbations and generate a pan-lung disease target spectrum. Cis-expression quantitative trait loci (cis-eQTL) data from lung tissue and whole blood were integrated with Mendelian randomization (MR) and colocalization analyses to prioritize causal genes. An independent validation cohort confirmed reproducibility.

Results

The in silico treatment analysis identified a therapeutic target spectrum of 3570 genes across 12 lung traits. Of these traits, three—lung cancer, FVC, and the FEV1/FVC ratio—had genes with causal relationships established. Specifically, 2 genes were linked to lung cancer, 8 to FVC, and 20 to the FEV1/FVC ratio. These findings were validated via colocalization testing and confirmed in an independent validation set.

Conclusions

This study identified 30 putative causal therapeutic targets for lung cancer and lung function traits, offering a systematic and robust foundation for the development of targeted therapies for lung diseases.