Žiga Avsec

AlphaGenome Atlas: in silico mutagenesis of the entire human genome improves prioritization and interpretation of non-coding variants

AlphaGenome Atlas predicts regulatory effects of non-coding variants across the human genome, improving their prioritization and interpretation.

An encyclopedia of enhancer-gene regulatory interactions in the human genome

ENCODE-rE2G, a classifier trained on K562 CRISPR perturbations, maps 13,455,443 enhancer-gene interactions across 352 biosamples, matching or exceeding prior models on CRISPR, eQTL, and GWAS benchmarks.

Effective gene expression prediction from sequence by integrating long-range interactions

Enformer, a sequence-based transformer integrating elements up to 100 kb away, lifts held-out gene expression prediction (correlation 0.81 to 0.85 versus Basenji2) and variant effect prediction.

Predicting Mean Ribosome Load for 5'UTR of any length using Deep Learning

The paper introduces a deep learning model for predicting mean ribosome load in 5'UTRs of any length, showing improved generalization and variant interpretation capabilities.

Base-resolution models of transcription factor binding reveal soft motif syntax

The paper introduces BPNet, a deep learning model that predicts base-resolution ChIP-nexus binding profiles of TFs and identifies soft syntax rules for cooperative TF binding.

Query to reference single-cell integration with transfer learning

This paper introduces scArches, a deep learning method for mapping query datasets onto reference single-cell atlases using transfer learning, effectively reducing batch effects and preserving biological information with fewer parameters.

Modular modeling improves the predictions of genetic variant effects on splicing

The paper introduces MMSplice, a modular framework for predicting genetic variant effects on splicing, achieving performance comparable to state-of-the-art methods.

OUTRIDER: A statistical method for detecting aberrantly expressed genes in RNA sequencing data

OUTRIDER detects aberrantly expressed genes in RNA-seq data by modeling co-variation and using significance-based thresholds to identify outliers.

Kipoi: accelerating the community exchange and reuse of predictive models for genomics

Kipoi provides a framework for standardized, reusable machine learning models in genomics, enabling tasks like variant effect prediction and model benchmarking.

Modeling positional effects of regulatory sequences with spline transformations increases prediction accuracy of deep neural networks

This paper introduces a spline transformation module for deep neural networks to model distances in regulatory sequences, improving prediction accuracy for RNA-binding protein binding sites and splice branchpoints.

Cis-regulatory elements explain most of the mRNA stability variation across genes in yeast

This study quantifies how mRNA sequence features influence stability in yeast, revealing key motifs and dependencies on degradation pathways.