Population-level allelic dispersion modeling by maelstRom yields genome-wide maps of allele-specific dysregulation during early carcinogenesis

TL;DR

This paper introduces a new method called maelstRom to detect early gene dysregulation in cancer by analyzing allele-specific expression differences.

Contribution

The paper introduces differential allelic dispersion (AD) as a novel measure for detecting early allele-specific dysregulation in cancer.

Findings

maelstRom successfully benchmarks using known copy number alterations in renal cancer data.

Increased AD is observed in loci with random monoallelic expression, such as the X chromosome.

Early dysregulated genes like FBP1 and pathways like the pentose phosphate pathway are identified in renal cancer.

Abstract

Since its inception, RNA sequencing has been pivotal in studying differential gene expression. Despite its extensive results in large-scale oncological studies, differential expression predominantly reflects a response to cancer. Therefore, we introduce differential allelic dispersion (AD) as a more effective measure. AD highlights consistent differences in expression between the 2 alleles of a gene that is, unlike cis-expression quantitative trait loci, independent of normal genetic variation. Such differences can, for example, arise from prevalent copy number alterations or epimutations occurring in the original cancer cell, which are mitotically expanded during cancer growth, making increased AD a marker for allele-specific dysregulation in early carcinogenesis.

We present the maelstRom R/C++ software package that enables (differential) AD analysis solely requiring large-scale RNA sequencing data. Using the The Cancer Genome Atlas renal clear cell carcinoma cohort as a case study, we successfully benchmark maelstRom’s AD modeling using known copy number alterations. We also detect increased AD for loci featuring normal random monoallelic expression, including the X chromosome, but demonstrate minimal interference with cancer-specific AD detection. Finally, we identify early dysregulated genes (e.g., FBP1, CCDC8, ECHS1, CLDN7) and pathways in renal cancer, often related to metabolism (e.g., pentose phosphate pathway). Strikingly, many of these genes are known causal contributors to renal carcinogenesis.

Differential AD clearly indicates early dysregulation in renal cancer, complementing basic differential expression analysis in cancer transcriptomics. AD is also relevant to study random monoallelic expression and may equally detect allele-specific (dys)regulation during early development or in noncancer diseases. maelstRom is available as an open-source software package at github.com/Biobix/maelstRom.

Graphical Abstract

Figures