In the field of metagenomics, the classification of long reads is a rapidly developing topic. Today’s tools perform well but fail to classify all reads. With this thesis work, we set ourselves the goal of improving the performance of Kraken2, one of the most used and accurate classifiers. To this end it was decided to experiment with the LabelPropagation technique already presented in ClassGraph, combined with RefineLabel techniques, on a graph, starting from the results of kraken2. The tool produced was then evaluated under different conditions, on different datasets, showing in all cases that it was able to improve performance.
Nell’ambito della metagenomica la classificazione delle long read è un tema in forte sviluppo. I tool odierni hanno buone performance, ma non riescono a classificare tutte le reads. Con questo lavoro di tesi ci si è posti l’obiettivo di migliorare le prestazioni di Kraken2, uno dei classificatori più utilizzati e accurati. A tal fine si è deciso di sperimentare la tecnica di LabelPropagation già presentata in ClassGraph, combinata a delle tecniche di RefineLabel, su grafo, a partire dai risultati di kraken2. Il tool prodotto è stato poi valutato a diverse condizioni, su diversi dataset, mostrando nella totalità dei casi di riuscire a migliorare le prestazioni.
Metagenomic Classification of Long reads with overlap graphs
LUCIANI, MATTIA
2021/2022
Abstract
In the field of metagenomics, the classification of long reads is a rapidly developing topic. Today’s tools perform well but fail to classify all reads. With this thesis work, we set ourselves the goal of improving the performance of Kraken2, one of the most used and accurate classifiers. To this end it was decided to experiment with the LabelPropagation technique already presented in ClassGraph, combined with RefineLabel techniques, on a graph, starting from the results of kraken2. The tool produced was then evaluated under different conditions, on different datasets, showing in all cases that it was able to improve performance.File | Dimensione | Formato | |
---|---|---|---|
Luciani_Mattia.pdf
accesso riservato
Dimensione
2.63 MB
Formato
Adobe PDF
|
2.63 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/29230