Artur Szałata

This paper introduces PertOmni, a multimodal framework that aligns transcriptomic data with text and image embeddings to improve perturbation effect prediction.

CellFlow predicts single-cell phenotypes from complex perturbations using flow matching, demonstrating accuracy in cytokine stimulation, drug treatments, and developmental modeling.

MOBER integrates transcriptomic data from cancer cell lines, xenografts, and clinical tumors using deep learning to remove batch effects and improve clinical translatability.

This paper introduces a VAE-based model for integrating multimodal single-cell data, showing it can map queries across modalities and improve data imputation and biological discovery.

This paper integrates 36 single-cell transcriptomics datasets from human neural organoids into a unified atlas to assess cell type coverage and compare organoid protocols with primary brain tissue.