Petr Šimeček

This study uses ESM3 to generate knotted proteins, finding high success rates with guided decoding and robustness in real knotted proteins.

The paper evaluates genomic language models as DNA compressors, finds smaller vocabularies perform better, and maps information content across the human genome.

The paper compares AlphaFold2, ESMFold, and OmegaFold on new PDB structures, finding AlphaFold2 most accurate but slower, while ESMFold and OmegaFold offer faster predictions with varying strengths in specific cases.

This study uses machine learning to predict protein knotting from amino acid sequences, achieving 98.5% accuracy using AlphaFold predictions and UniProt data.

The paper introduces a collection of genomic sequence classification datasets and a baseline CNN model, distributed as a Python package to improve reproducibility in genomics machine learning.

This study evaluates Transformer models for genomic sequence classification, comparing pre-trained and from-scratch approaches on benchmark datasets.

This study uses 16S rRNA gene sequencing in Diversity Outbred mice to identify genetic variants affecting gut microbiome composition, finding significant heritability and novel QTLs related to immune and metabolic pathways.

PENGUINN uses Convolutional Neural Networks to predict G-quadruplex forming sequences, outperforming existing methods.

The paper identifies and characterizes noncrossovers in mice using NGS sequencing, finding they differ in distribution from crossovers and suggest sister chromatid repair.

R/qtl2 is a software tool for QTL mapping in multi-parent populations and high-dimensional data, offering features like linear mixed models and SNP imputation.

The paper generates high-resolution genetic maps of mouse inbred strains and identifies genetic variations and gene conversions.

This study examines how sex and diet influence physiology and liver gene expression in Diversity Outbred mice, identifying genetic and sex-by-diet interactions affecting metabolic traits.