Dmitry Penzar

The paper benchmarks methods for inferring transcription factor binding specificities using a large community challenge and finds deep learning models outperform traditional methods, though positional weight matrices are competitive.

EPIC challenges teams to predict transcription initiation from DNA sequence in non-model metazoans, aiming to assess generalizability beyond human and model species.

This study identifies DNA binding motifs for 98 human transcription factors using meSMiLE-seq, revealing methylation-sensitive binding preferences for some TFs.

The paper uses GHT-SELEX to show many human TFs have high intrinsic sequence specificity and complex DNA binding, often matching in vivo ChIP-seq results.

ArChIPelago combines multiple PWMs using classic machine learning to improve TFBS prediction, outperforming individual PWMs and other models across species.

The paper evaluates cross-platform DNA motif discovery tools using 4,237 experiments to identify reliable transcription factor binding motifs and highlights challenges in motif detection.

GAME provides a standardized framework for benchmarking genomics models across tasks, enabling consistent evaluation and community contributions.

The paper identifies functional cis-regulatory elements in three human cell types using lentiMPRAs and develops predictive models for CRE function.

This paper introduces PARADE, a generative AI framework for designing mRNA UTRs with cell-type specificity, showing improved therapeutic efficacy in preclinical models.

GENA-LM introduces transformer-based DNA language models for long sequences, capable of handling up to 36,000 base pairs and supporting multispecies and taxon-specific applications.

The paper systematically determines the sequence specificity of 332 poorly characterized human transcription factors, identifying over 4,000 experiments and revealing tens of thousands of new TF binding sites.

The paper evaluates and optimizes sequence-based gene regulatory deep learning models using a DREAM Challenge dataset, showing improvements through architectural and training strategy refinements.

GENA-Web is a web platform using DNA language models to annotate genomic elements, including promoter activity, splice sites, chromatin features, and enhancer activity in Drosophila.

The paper introduces RibonanzaNet, a deep learning model trained on crowdsourced RNA chemical mapping data to predict RNA structure and related properties.

LegNet is a deep learning model for predicting gene expression from DNA sequences, outperforming existing models and enabling promoter sequence design.

The paper identifies allele-specific transcription factor binding events across the human genome using ChIP-Seq data, revealing regulatory variants linked to disease and molecular mechanisms.