Bianca-Maria Cosma

The paper predicts other meta-omics data from metagenomics using machine learning, showing reliable predictions for transcript and metabolite abundances but challenges in protein profiles, and demonstrates utility for inflammatory bowel disease classification.

SAP improves bacterial gene function prediction using protein embeddings and conserved synteny, identifying novel toxins in enterococci.

The paper benchmarks long-read de novo assemblers for eukaryotic genomes, finding that Flye performs best overall, and longer reads can improve assembly quality depending on genome complexity.