Whole Genome Sequencing (WGS) Pipeline

Each WGS job receives a dedicated analysis environment. Files are transferred over a private network, analyzed in isolation, and results are encrypted at rest.

Same 3-layer validation as the 16S pipeline:

• Layer 1 (upload): GZIP magic bytes, FASTQ format, FASTA detection, min size check
• Layer 2 (pre-pipeline): Preflight tool check — verifies all required tools are installed before starting
• Layer 3 (post-analysis): Assembly quality metrics (N50, contig count) evaluated in the report

Fastp v1.1.0

All-in-one FASTQ preprocessing: adapter removal, quality trimming, read filtering, and QC reporting.

Output: Trimmed FASTQ, HTML QC report, JSON stats (total reads, GC content, quality distribution, adapter rates)

Kraken2 v2.17.1 Bracken v2.9

Kraken2 — k-mer Classification

Classifies each read by matching k-mers (k=35) against a pre-built database. Uses a lowest common ancestor (LCA) algorithm to handle ambiguous matches.

Bracken — Bayesian Abundance Re-estimation

Refines Kraken2's output using Bayesian re-estimation of species abundance, redistributing reads assigned to higher taxonomic levels.

Output: Primary species, confidence %, full taxonomy report, species distribution

SPAdes v4.2.0 QUAST v5.3.0

SPAdes — De Novo Assembly

Multi-k-mer de Bruijn graph assembler optimized for bacterial isolates. Uses the --isolate mode for single-genome assembly.

QUAST — Assembly Quality Assessment

Evaluates assembly quality without a reference genome:

• N50 — 50% of assembly contained in contigs ≥ this length (higher = better)
• Contig count — fewer = better assembly (ideal: <100)
• Total length — should match expected genome size (±10%)
• GC content — validates against known species GC%
• Largest contig — indicates maximum contiguity achieved

Output: Scaffolds FASTA, QUAST report (TSV + HTML), assembly stats

Bakta v1.12.0

Rapid, comprehensive annotation of bacterial genomes. Bakta uses a hierarchical annotation approach with multiple databases.

Output: GFF3 annotation, protein FASTA (.faa), nucleotide FASTA (.ffn), GenBank format, annotation summary

AMRFinder+ v4.2.7 (NCBI)

Identifies antimicrobial resistance genes, stress response genes, and virulence factors. Uses both nucleotide and protein-level searches.

Output Categories

• AMR genes — acquired resistance determinants (e.g., mecA, vanA, blaCTX-M)
• Point mutations — chromosomal mutations conferring resistance (e.g., gyrA S83L)
• Stress response — metal/biocide resistance genes
• Virulence factors — pathogenicity-associated genes

mlst v2.33.1 (Torsten Seemann)

Multi-Locus Sequence Typing identifies the sequence type (ST) of the isolate by matching 7 housekeeping gene alleles against PubMLST schemas.

• Scheme auto-detection — selects the correct MLST scheme based on species
• Allele matching — exact match against >130 species-specific schemas
• Novel alleles — flagged with ~ prefix if closest match is inexact

Output: Scheme name, sequence type (ST), 7 allele profiles

matplotlib v3.9 WeasyPrint v63.1 Gemini AI

Charts Generated

• QC before/after — read count and quality comparison
• Species distribution — top species by abundance (pie/bar)
• Assembly quality — contig size distribution
• Gene composition — CDS, rRNA, tRNA breakdown
• AMR summary — drug classes and resistance mechanisms

AI Interpretation

Google Gemini provides section-specific interpretations for Species, Assembly, AMR, MLST, and QC — translating technical metrics into plain-language clinical context.

Deliverables

• Interactive Web Report — full results with expandable AI cards
• PDF Report — publication-ready with all figures
• Research Data ZIP — assembly FASTA, annotation GFF3/FAA/FFN, Kraken report, QUAST TSV, AMR results, fastp JSON

Reference Databases