Lila Kari A

The paper identifies DNA sequence patterns associated with thermal adaptation in extremophiles, independent of phylogeny, using structural topic modelling of short sequence motifs.

This study investigates the pervasiveness and variation of genomic signatures across eukaryotic genomes using k-mer-based FCGR analysis, identifying localized exceptions and developing tools for improved taxonomic classification.

The paper identifies 15 bacterium-archaeon pairs with similar genomic signatures despite maximal taxonomic divergence, showing environmental pressures drive convergent genome-wide patterns.

This study shows that environmental factors like temperature and pH leave a detectable genomic signature in extremophiles, using machine learning to analyze k-mer frequencies across genomes.

iDeLUCS is a deep learning tool for clustering DNA sequences without alignment or taxonomy, outperforming existing methods on real and synthetic datasets.

SomaticSiMu simulates mutational signatures in genomic sequences and evaluates its reliability using machine learning and mutational signature extraction.

DeLUCS clusters DNA sequences without alignment or homology, achieving high accuracy on diverse datasets and outperforming traditional methods.

MT-MAG is a machine learning tool for taxonomic classification of MAGs, achieving high accuracy and providing interpretable results for complete or partial classifications.

This paper uses machine learning and genomic signatures to classify novel pathogens, demonstrating high accuracy in identifying the origin and taxonomy of the COVID-19 virus.

MLDSP-GUI is an open-source, alignment-free tool for DNA sequence comparison and analysis, offering taxonomic and evolutionary applications.

ML-DSP uses machine learning and digital signal processing to classify genomes quickly and accurately across all taxonomic levels, outperforming existing methods in speed and scalability.

The paper introduces an open-source, alignment-free k-mer based tool for accurate and fast HIV-1 subtype classification, achieving 97% accuracy and outperforming existing methods in speed and privacy.