M. Sohel Rahman

This paper introduces MirrorDNA, a DNA language model with a strand-equivariant tokenizer that improves prediction consistency across DNA strands.

The paper introduces a multi-modal framework for biomarker discovery in TNBC using public data, achieving high predictive performance and generating testable hypotheses.

This paper models tissue-specific biological age using machine learning on GTEx data, finding that extreme aging in specific tissues correlates with increased mortality risk.

This paper introduces scPSS, a method to quantify pathological shifts in single-cell data by comparing diseased cells to healthy references, showing it can predict individual health conditions.

The paper introduces a framework to model feature interactions and temporal dynamics in Alzheimer's disease progression using ADNI data, identifying key clinical and molecular markers.

CRISMER improves CRISPR-Cas9 off-target prediction and optimization using a hybrid deep learning model, achieving higher accuracy and fewer parameters than existing methods.

PRIEST predicts SARS-CoV-2 mutations with immune escape potential using temporal evolutionary data and deep learning.

The paper introduces CRISPR-DIPOFF, a deep learning model for CRISPR-Cas9 off-target prediction that improves performance and provides interpretability through integrated gradients and genetic algorithm optimization.

CC-Tempo improves prediction of cell fate by integrating cell-cell communication scores into a temporal model of cellular differentiation.

DeepBend is a convolutional neural network that identifies DNA bendability motifs and their spatial patterns, revealing known and novel motifs and their role in chromatin structure.

The paper proposes deep learning models for predicting succinylated lysine residues using biochemical and physicochemical properties, achieving improved sensitivity and specificity through threshold tuning.

QT-GILD improves species tree estimation by imputing missing quartets from incomplete gene trees using deep learning, enhancing accuracy despite data gaps.

The paper analyzes how EBNA2 and EBNA3 vTFs bind the human genome across cell-types, showing they use both cell-type specific and independent cofactors and reveal novel sequence motifs.

SICaRiO improves indel detection by using a gradient boosting classifier trained on GIAB data, enhancing variant calling pipelines and identifying genomic features responsible for indel calling errors.

The paper develops a pipeline to classify SARS-CoV-2 genome sequences for virulence and predict mutations using deep learning.

The paper analyzes missense variants in 1,330 disease genes using 3D protein structures to identify mutational hotspots and their functional impact.

NORTH is a Naive Bayes-based algorithm for orthologous gene clustering that achieves high accuracy and scalability across diverse organisms and large datasets.

The paper presents upper bounds and algorithms for sorting binary and ternary strings using prefix block-interchange operations and provides a classification for ternary strings based on required operations.

CRISPRpred(SEQ) is a sequence-based tool that outperforms DeepCRISPR in predicting sgRNA on-target activity for three out of four cell lines using traditional machine learning.

CHilEnPred uses a CNN with Hilbert Curve DNA sequence representation to predict enhancers, achieving 94.97% accuracy on FANTOM5 data.

The paper investigates whether multi-objective MSA formulations improve phylogenetic tree estimation and finds they do, outperforming single-objective methods.

The paper revisits a 2-approximation algorithm for sorting by prefix transpositions, improving its running time to O(n log n) using a permutation tree data structure.