VAPOR

VAPOR applies a graph-based classifier to classify Influenza virus samples from raw short-read sequencing data by identifying closest reference sequences to select optimal mapping references and validate assemblies.

Key Features:

• Graph-based classifier: Implements a graph-based classifier to compare sample reads against candidate reference sequences.
• Reference selection: Accepts a fasta file containing full-length sequences (exceeding 20,000 base pairs) for a specific segment alongside a set of reads and identifies the reference sequence most closely matching the sample strain.
• Assembly validation: Validates assemblies and addresses misassembly issues arising from de novo assembly processes.
• Non-human strain detection: Detects non-human origin strains to support identification of inter-species transmission events.
• High-identity retrieval: Retrieved references with an average identity of over 99.8% to assemblies when tested on real whole-genome sequencing samples.
• Increased mapping: Increases the proportion of mapped reads by up to 13.3% compared to standard human reference viruses, reducing read loss during mapping.
• Robustness to RNA variation: Accounts for rapid RNA genome evolution, intra-host variation, and noise in short-read data.
• Extensibility: Has potential to be adapted for classification and reference selection of other RNA viruses.

Scientific Applications:

• Influenza surveillance and research: Improves reference selection and assembly validation in influenza genomic surveillance and research workflows.
• Mapping pipeline enhancement: Augments mapping pipelines by increasing mapped reads and mitigating read loss during mapping.
• Zoonotic transmission detection: Aids detection of inter-species transmission by identifying non-human origin strains.
• De novo assembly QC: Provides validation and quality control for de novo assembled influenza genomes.
• Application to other RNA viruses: Can be applied to classification and reference selection tasks for other RNA viruses.

Methodology:

Uses a graph-based classifier; inputs a fasta file of full-length sequences (>20,000 base pairs) for a specified segment together with short-read sequencing reads; identifies the reference sequence that most closely matches the sample strain.