Maximilian Sprang

Prot2Prop predicts six protein developability properties using a multitask framework with a frozen ProstT5 encoder and task-specific adapters, achieving strong performance on both classification and regression tasks.

This study investigates age-related changes in dendritic cells across tissues, finding tissue-specific alterations in DC development and function, with preserved cell-intrinsic capabilities in aged bone marrow.

The paper introduces MIRNAPEX, a framework that accounts for alternative polyadenylation to better understand miRNA differential expression by integrating target-gene expression and 3'UTR isoform usage.

The paper uses a multi-property fine-tuned language model to predict lipid retention times from peptide data, improving accuracy in data-sparse scenarios.

The paper re-evaluates the identity of CD4+ CD56+ CD16+ cells in human PBMC, finding they resemble NKT cells rather than exTreg, and explores Treg stability and Tr17 induction for therapeutic applications.

This paper reviews 86 studies on biologically informed neural networks for multi-omics integration, highlighting trends and challenges in terminology, tools, and evaluation methods.

This study uses machine learning to predict miRNA expression and identify genetic regulators, revealing miRNA-gene regulatory relationships and pathway-specific associations.

The paper improves seqQscorer by integrating the ENCODE blocklist to enhance ChIP-seq quality assessment accuracy and efficiency.

This study shows that transformer models struggle to capture DNA's functional complexity, suggesting inherent limitations that scale cannot overcome.

The paper reduces CRISPR dark matter by analyzing spacers in antibiotic-resistant bacterial genomes and links CRISPR-Cas systems to membrane proteins involved in phage defense.

This paper develops machine learning models for automated quality control of next generation sequencing data, validated on internal and external datasets.