Decoding sequence determinants of gene expression in diverse cellular and disease states.

Cell-type-specific gene expression prediction from DNA sequence

Decima is a deep learning model trained on over 22 million single-cell and single-nucleus RNA-seq cells to predict cell-type- and condition-specific gene expression from surrounding DNA sequence. Unlike prior models trained on bulk expression from healthy tissues, Decima captures the regulatory properties of specific cell types and disease states. The tool demonstrates its ability to interpret noncoding variants at cell-type resolution, paving the way for improved annotation of regulatory variants identified in clinical WGS.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

Interpreting noncoding variants is one of the major challenges in modern diagnostic genomics, where a growing proportion of pathogenic variants lie in regulatory regions (promoters, enhancers, deep splicing sites). Published in Nature Methods with a training base of 22 million cells, Decima represents a significant advance for clinical WGS laboratories seeking to reclassify VUS in noncoding regions.

Clinical impact: 2/3 · Evidence strength: 2/3 · Novelty: 2/2 · Sample size: 1/1 · Journal quality: 1/1 → Total: 8/10