Amin Emad

DiffDose uses automatic differentiation to optimize drug dosing regimens by turning mechanistic PK/PD and QSP models into gradient-based tools for personalized treatment design.

The paper improves sequence-to-expression models by incorporating chromatin accessibility, showing better performance in predicting gene expression and cell type-specific patterns.

This paper introduces scLSD, a thermodynamics-inspired model for cell differentiation that reconstructs differentiation trajectories and predicts fate commitment, outperforming existing methods.

CellPace simulates, interpolates, and forecasts single-cell developmental dynamics using a temporal diffusion model with continuous time encoding, achieving state-of-the-art performance on mouse data.

This paper identifies data leakage in pre-trained pLLMs used for protein-protein interaction inference and shows their limitations in generalization.

The paper introduces GRouNdGAN-Toolkit, a toolkit for simulating scRNA-seq data and benchmarking GRN inference methods using causal generative models.

CLASP integrates protein sequence, structure, and text-based descriptions using geometric deep learning and contrastive learning to create embeddings that improve zero-shot classification and clustering.

The paper introduces Mach-1, an RNA foundation model that predicts transcriptome architecture from DNA sequence, validated through experimental perturbations and CRISPR editing.

INTREPPPID improves cross-species protein-protein interaction prediction using orthologue-informed neural networks and outperforms existing methods on evaluation datasets.

GRouNdGAN simulates single-cell RNA-seq data using causal generative adversarial networks guided by gene regulatory networks, outperforming existing simulators and enabling in-silico perturbation experiments.

This study evaluates interpretable deep learning models for drug response prediction, finding that explicit pathway integration reduces accuracy and that black-box models often perform best.

MARSY predicts drug combination synergy scores using a multitask deep learning framework, improving prediction performance over existing methods and providing new predictions for 133,722 drug-pair combinations.

RAPPPID improves protein-protein interaction prediction by using regularised AWD-LSTM twin networks, outperforming existing methods on unseen proteins.

The paper introduces two deep learning models for drug response prediction that use bipartite graphs to improve accuracy by incorporating drug structure and cell line relationships.

This paper develops a deep learning framework called TINDL to predict cancer patient drug response and identify drug sensitivity biomarkers using preclinical samples, showing improved performance over other models.

This study identifies transcriptional regulatory networks and kinases involved in SARS-CoV-2 infection response, highlighting potential therapeutic targets for COVID-19.

This paper evaluates machine learning algorithms for predicting clinical drug response using preclinical samples and introduces TG-LASSO, which improves prediction accuracy for 7 out of 13 drugs.

This study introduces InPheRNo, a method to reconstruct phenotype-relevant transcriptional regulatory networks, showing it can identify cancer-specific regulatory mechanisms and prognostic TFs.