Philip T. L. C. Clausen

A ProtT5-embedding, ConvNext-plus-attention ensemble predicts human pathogenic capacity from whole bacterial genomes, outperforming prior predictors on species absent from training and flagging influential proteins.

This paper develops a random forest model to predict plasmid hosts, achieving 0.662 and 0.867 Matthews correlation coefficients at species and order levels, respectively.

The paper evaluates metagenomic software tools using realistic datasets, highlighting improvements in assembly and taxon profiling while identifying remaining challenges in virus/archaea detection and reproducibility.

MINTyper is a SNP typing method that clusters DNA isolates using a mix of short and long reads, showing accurate clustering even with lower-quality long reads.

CCMetagen improves metagenomic classification by using weighted read mappings, outperforming existing tools in precision and F1 scores for species-level identification.