Abdul Muntakim Rafi

gRely provides reliability scores for sequence-to-expression models, improving prediction filtering beyond magnitude thresholds in low-magnitude regimes.

This study evaluates active learning strategies for sequence-to-expression models, finding that uncertainty-based methods outperform random sampling and identify informative sequences with distinct biological features.

Yorzoi predicts RNA-seq coverage for yeast DNA sequences, enabling in-silico testing and design of synthetic DNA constructs.

This paper identifies homology-induced data leakage in genome-trained sequence models and introduces hashFrag to improve evaluation and generalizability.

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.

The paper quantifies the gene regulatory activity of naive DNA in yeast and human cells, finding that naive DNA can produce transcripts and chromatin marks similar to evolved sequences, suggesting biochemical activity is the default state.

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

GIL designs custom indexing primers for multiplexed sequencing, offering customization and compatibility with existing primers.