Bioinformatics Evidence

Bioinformatics/Coding Plan: Long-read / Hybrid Assemblies for Actionable ARG Surveillance

• Short-read assemblies often fragment plasmids/MGEs, which degrades reconstructability and linkage confidence (especially across taxa with poor database representation). short-read contiguity/database representation limit plasmid reconstruction
• Long-read / hybrid approaches improve structural resolution, which supports resolving ARG genetic context and enabling ARG–plasmid–host linkage required for higher-confidence surveillance prioritization. long reads provide high-resolution contextualization
• In large surveillance settings, long-read / hybrid subsets are used to update and validate plasmid-level inferences (e.g., tracking blaKPC plasmids over time). long-read/hybrid used for plasmid tracking
• Even when short-read WGS can detect mobile elements at scale, long-read is recommended to resolve complete architectures when contigs are fragmented. short-read fragmentation motivates long-read architecture confirmation

3) Concrete analysis plan (what you code/compute)

• Define the objective schema: for each sample, record ARG presence, MGE/plasmid candidates, and a host taxon confidence.
• Implement controls for assembly integrity and contamination screening; if viability/internal standards are part of your surveillance design, store them as explicit covariates because the mobility-risk framework anticipates such integration. viability/internal standards improve interpretability in surveillance

• Detect acquired ARGs and resistance-associated mutations using your preferred classifier/detector that targets known resistance determinants (e.g., CARD-based). Note: read-based methods may not discover novel ARG families. read-based ARG calling depends on CARD (novelty limitation)
• Detect mobile genetic elements / plasmid replicons and build a mobility proxy score (e.g., ARGs colocated with MGE signatures), but treat it as proxy-only until upgraded with contextual linkage.
• Because plasmid detection/reconstruction from short reads is accuracy-limited by contiguity and database representation, explicitly track expected uncertainty for downstream prioritization. short-read plasmid inference uncertainty depends on contiguity and database coverage

• Compute a priority score driven by: (i) ARG clinical risk rank (domain-specific), (ii) evidence of mobility proxy, (iii) diversity/serotype/species novelty, and (iv) ambiguity level (e.g., ARG detected without any consistent plasmid/MGE contiguity).
• Upgrade only top-K samples per surveillance interval to conserve budget, consistent with the tiered framework. priority follow-up saves cost

• Perform long-read or hybrid assembly and then annotate: ARGs, plasmid/MGE signatures, and replication/transfer loci. The mobility-risk integration explicitly lists long-read metagenomics/hybrid assembly and MGE detection as components for contextualization. long-read/hybrid + MGE detection for contextual linkage
• Implement contiguity-aware confidence: only call ARG–plasmid linkage when the assembly provides adequate genetic context spanning ARG + plasmid/replication/transfer features (otherwise mark linkage as uncertain/partial).
• Use host resolution logic: when assemblies allow reliable taxonomic assignment of the contig(s) carrying the linked ARG/MGE features, store a host taxon with a confidence flag.

• For the most consequential cases, validate linkage using orthogonal methods noted in the mobility-risk framework (e.g., ddPCR linkage or Hi-C/3C association). ddPCR and Hi-C/3C validate ARG linkage

4) Example surveillance metrics to implement (from published surveillance papers)

• Database bias: short-read plasmid detection/reconstruction accuracy depends on plasmid representation in databases; novelty and taxa outside coverage will be systematically underperforming. database representation limits plasmid detection
• Architectural fragmentation: short-read contigs may not resolve complete phage-plasmid structures; long-read confirmation can be necessary. fragmentation motivates long-read confirmation
• ARG genotype calling pitfalls: interrupted genes and multicopy/allelic diversity can cause genotype–phenotype discordance; incorporate gene integrity/copy-number aware logic rather than assuming “hit == functional gene.” gene interruptions/copy number affect ARG genotyping accuracy
• Mobility ≠ epidemiological outcome: the surveillance framework explicitly treats mobility as a proxy for dissemination risk; validate with independent clinical/environmental outcomes where possible, and treat improvements in prediction as hypotheses to test. mobility is a proxy; validate risk gains