Basak Eraslan

Predicting differentially expressed genes and expression shifts per condition from database embeddings plus per-gene evidence, PIE reaches 1.2–3.2 times the best baseline's AUPRC on unseen contexts, perturbations and datasets.

State, a set-based transformer trained on over 100 million perturbed cells, predicts perturbation responses in new cellular contexts better than mean, linear and deep learning baselines.

The paper introduces a framework for multimodal chemical screens combining RNA, protein, and epigenomic readouts, revealing compound-specific mechanisms and off-target effects in T cells.

The paper identifies urinary protein signatures linked to LRRK2 dysfunction in Parkinson's disease using multi-cohort proteomics and machine learning, showing cross-species relevance and potential as pharmacodynamic markers.

Compressed Perturb-seq reduces cost and complexity of CRISPR screens by using random composite perturbations and computational decomposition, achieving similar accuracy to conventional methods with greater power to detect genetic interactions.

The paper uses Perturb-seq to systematically study 1,130 E3 ligases and their roles in inflammatory responses in dendritic cells, revealing co-functional modules and genetic interactions.

The paper quantifies protein-to-mRNA (PTR) ratios across 29 human tissues and identifies sequence determinants influencing these ratios, finding many new post-transcriptional regulatory elements.

The paper creates a proteome and transcriptome atlas of 29 human tissues, revealing differences in mRNA-protein abundance and challenges in proteogenomics.