Fran Supek

The paper identifies a mutational signature in normal kidney cells linked to kidney cancer initiation through hypoxia-driven genetic changes.

This study quantifies sequence determinants of human programmed stop codon readthrough using deep mutational scanning and identifies gene-specific and context-dependent effects.

The paper develops MutFormer, a transformer model to predict non-coding cancer driver mutations at base-pair resolution using DNA sequence and mutational signatures.

This study maps the mutational consequences of MMR gene deficiencies in cancer cells using CRISPR, mutation accumulation, and whole-genome sequencing, revealing multidimensional mutational phenotypes.

The paper develops a model to predict nonsense-mediated mRNA decay (NMD) across human genes using genomic data and mutational scanning, showing improved accuracy and revealing quantitative rules of NMD implementation.

The paper develops ORIFormer and MuSAS to predict and map human replication origins, revealing DNA sequence determinants and tissue-specific regulation through chromatin accessibility.

The paper quantifies NMD efficiency across human tissues and individuals, identifies tissue-specific and inter-individual variability, and links NMD efficiency to genetic factors and clinical outcomes.

The paper characterizes somatic mutations in human tRNA genes, showing they are mutagenesis hotspots with tumor-specific loads and can generate altered tRNA variants affecting the genetic code.

DiffInvex identifies chemotherapy resistance genes by analyzing 10,000 cancer genomes, finding known and novel genes under conditional selection for specific therapies.

The paper identifies copy number-independent allelic imbalance in mRNA as a selected feature in cancer with prognostic significance.

The paper quantifies and predicts pathogenic stop codon readthrough by small molecules across ~5,800 human PTCs using 8 drugs, revealing drug-specific readthrough patterns and building predictive models.

The paper identifies common driver events in human cancer by analyzing copy number alterations and their interactions with mutations in oncogenes and tumor suppressor genes.

The paper identifies sub-gene scale mutation rate heterogeneity linked to DNA hypomethylation and different mutational mechanisms across gene regions.

The paper identifies two global RMD signatures of mutation rate redistribution in tumors associated with RB1 and TP53 pathway alterations and cell cycle changes.

This study examines mutational profiles in human cells exposed to proton and alpha radiation, finding subtle changes in mutation spectra and distributions but not overall increased mutation rates.

The paper estimates the prevalence of TP53-loss phenocopying events in tumors and identifies new genes like USP28 and CNOT2 that contribute to p53 pathway inactivation without TP53 mutations.

The paper identifies mutational signatures as markers of drug sensitivity in cancer cells by analyzing drug and genetic screening data.

This paper introduces a framework for mutational signature analysis using DNA shape parameters to better understand mutagenesis mechanisms and improve mutation rate baselines.

This study identifies 207 associations between rare germline variants and somatic mutational processes in cancer genomes, revealing novel genes influencing DNA repair and mutation signatures.

The paper identifies mislabelled cancer cell lines using transcriptomic and epigenomic data and proposes a 'golden set' of accurately labelled cell lines for experimental use.

The paper identifies gene function patterns in genomic neighborhoods that predict gene function and phenotype, showing improved predictive power over existing methods.

The paper reports results from the CAFA3 challenge, showing improved computational protein function prediction and new annotations for over 1000 genes through experimental screens in C. albicans, P. aureginosa, and Drosophila.

The paper analyzes somatic retrotransposition events in 2,954 cancer genomes, identifying L1 insertions as significant contributors to structural variations and oncogenesis.

The paper shows overtraining in maximum likelihood species tree reconstruction leads to topologically incorrect trees, and identifies likelihood curve shape as a guide for optimal stopping.