María Rodríguez Martínez

Pop-Corn predicts how genetic perturbations reshape cell-type composition in single-cell and spatial contexts without reconstructing gene expression.

The paper develops a computational framework using optimal transport to analyze macrophage state transitions in atherosclerosis, identifying origin-dependent plasticity mechanisms and shared regulatory circuits in inflammation and fibrosis.

IMMREP25 evaluates TCR:pMHC binding prediction for unseen peptides, showing improved performance with structural modeling but still below random guessing.

The paper benchmarks generative methods for designing peptides in TCR-pMHC complexes, revealing method-dependent differences in backbone generation and conformational diversity.

The paper identifies chemoresistant cell populations in pediatric AML that improve risk prediction and reveal therapeutic targets for high-risk patients.

This paper introduces phenotype-driven data augmentation methods for transcriptomic data, showing improvements in patient stratification over existing approaches.

This study evaluates whether domain-specific protein language models outperform general models in immunology tasks, finding nuanced performance differences and highlighting key hyperparameters.

The paper uses a stochastic model to fit single-cell transcriptomic data and shows it can recapitulate mRNA distributions during B cell differentiation in germinal centers.

The paper identifies co-regulated transcription factors and BACH1 as a key driver in RA FLS using gene network analyses and experimental validation.

The paper identifies cell-specific gene regulators and networks in RA synovial tissues using RNA-seq data and computational methods, revealing key transcription factors and pathways involved in disease.

This study compares clone identification methods for B-cell repertoires and finds Shannon entropy to be the most robust diversity index, with alignment-free methods preferred for shorter reads.

The paper benchmarks 23 TCR-epitope prediction models, finding that using paired-chain alpha-beta and CDR1/2/V/J data improves performance, and simple distance-based methods can be competitive.

SurvBoard benchmarks multi-omics cancer survival models, showing statistical models outperform deep learning in calibration and that pan-cancer training improves performance.

The paper presents a stochastic hybrid model of DNA replication in fission yeast that incorporates protein mobility dynamics and experimental data to simulate full-genome duplication and replication timing.

PCfun annotates protein complex function using natural language processing and machine learning, integrating word embeddings from PubMed Central articles with supervised and unsupervised methods.

The paper investigates how genomic alterations affect molecular networks in prostate cancer by integrating genomic, transcriptomic, and proteomic data from 105 samples, identifying a sub-network predictive of tumor aggressiveness and recurrence.