Higher recall in metagenomic sequence classification exploiting overlapping reads.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Metagenomic sequencing enables advancements in ecology, medicine, and microbiology.
• Accurate taxonomic classification of metagenomic reads is a significant challenge.
• Current state-of-the-art methods achieve high precision but low recall (around 50%), classifying only half the reads.

Purpose of the Study:

• To develop a novel method for improving metagenomic read classification.
• To enhance the recall, f-measure, and species abundance estimation in metagenomic datasets.
• To address the limitations of existing classifiers in accurately assigning all reads.

Main Methods:

• Proposed a method named CLassification Improvement with Overlapping Reads (CLIOR).
• CLIOR leverages the information from the overlapping reads graph of the dataset.
• Applied CLIOR on top of the Clark-l classifier for evaluation.

Main Results:

• CLIOR substantially improves recall, sometimes doubling it, across simulated and synthetic metagenomes.
• On average, CLIOR increases recall while also improving precision on simulated datasets.
• On real metagenomes, CLIOR assigns more reads accurately, maintaining consistent abundance ratios.

Conclusions:

• CLIOR effectively boosts the recall of metagenomic classifiers by correcting or inferring read assignments.
• The method is versatile and can be applied to various classification algorithms.
• CLIOR is computationally efficient and can be run on standard hardware like a laptop.