/ publications / [3] predicting cancer evolution

Identifying driver genes and mutations:

Enrichments of somatic mutations indicate that ~1 in 5 synonymous mutations in oncogenes are cancer drivers. Involvement in known exonic splicing motifs and association to RNA-Seq data implicates many causal synonymous mutations to altered splicing. The 3’ UTRs of dosage-sensitive oncogenes also harbour causal mutations.

A statistical method, ALFRED, tests Knudson’s two-hit hypothesis to systematically identify inherited cancer predisposing genes // We identify novel genes, such as the chromatin modifier NSD1, which cause cancer through germline variants and somatic loss-of-heterozygosity // 1 in 50 tumors is associated with novel ALFRED genes

Classifying cell-of-origin of cancers via mutation patterns:

Density of somatic mutations across chromosomal domains is a mutational phenotype that can differentiate human tissues // Driver mutations are poor classifiers of cancer (sub)type, while passenger mutation-based phenotypes are highly predictive // Trinucleotide signatures and regional mutation density phenotypes are complementary in classifying tumors.

Hypermutation as a cancer vulnerability:

We identify HMCES, a protein linked to the protection of abasic sites, as a central protein for the tolerance of A3A expression. HMCES depletion results in synthetic lethality with A3A expression preferentially in a TP53-mutant background. Our results suggest that HMCES is an attractive target for selective treatment of A3A-expressing tumors.

"Prediction is very difficult, especially about the future." -- Niels Bohr.