Johannes Hingerl

Scooby-2 fine-tunes AlphaGenome to predict single-cell genomic tracks, improving correlation and variant effect prediction compared to Scooby and AlphaGenome.

The paper benchmarks variant effect scores using UK Biobank data on ultra-rare variant carriers, showing coding and splicing scores better predict protein abundance and traits than transcriptional regulation models.

Nona is a multimodal masking framework that unifies sequence-to-function models, language models, and generative models for functional genomics, improving predictions and revealing privacy vulnerabilities in ATAC-seq data.

scooby models scRNA-seq and scATAC-seq data from DNA sequence to capture cell-specific gene regulation and variant effects.

Flashzoi improves the Borzoi model by using rotary positional encodings and FlashAttention-2, achieving faster training and inference with reduced memory usage while maintaining accuracy in genomic analysis.

This paper introduces nucleotide dependency analysis to identify genomic functional elements, showing it outperforms existing methods in detecting regulatory elements and RNA structures.

This paper introduces species-aware DNA language models that capture regulatory elements and their evolution across 800 species, showing improved prediction and motif discovery.

This paper introduces Spectralis, a deep learning method for de novo peptide sequencing that improves sensitivity and precision in tandem mass spectrometry.