Output Interpretation¶

Every profiling run produces an HTML report. You can also request structured exports (JSON, PDF, TSV) with --export for automation and downstream processing.

Start with the HTML report

Use the HTML report first for interpretation context, then use JSON/TSV for automation and downstream processing.

HTML report (`*.report.html`)¶

The HTML report is a single, portable file you can open in any browser. It is always generated.

What to inspect first:

sample and project identity
total hits and matched rules
per-feature mutation details and consequences
phenotype and clinical phenotype context
optional manual classifications

Best for:

review by analysts and clinicians
sharing a portable report artifact

Detailed examples of important report tabs¶

Summary tab — Primary summary that condenses findings.

Database tab — Detailed overview of all mutations that have rule hits in the selected database.

Mutation tab — Detailed overview of all mutations, regardless of whether they have database hits.

JSON export (`*.results.json`)¶

A structured, machine-readable export for automation and reproducibility.

Top-level sections:

run
variant_result
coverage_gap
formula_rule_hit
sample_classification

Best for:

downstream pipelines
archival and deterministic regeneration
data integration with external systems

Reproducible artifacts

JSON exports are intended as reproducible artifacts and can be used directly with respro regenerate --json.

Regenerate a report from JSON:

respro regenerate \
  --project myrespro.db \
  --json my_output/sample_variants.results.json \
  --output my_output

See CLI Reference for all regeneration options.

TSV export (`*.results.tsv`)¶

A denormalized, tab-separated table of every annotated variant and its matched resistance rules. One row is emitted per (annotated variant × matched rule); variants with no matching rule still appear with empty rule columns. This mirrors the HTML Database Hits table but is flat and machine-readable.

Enable it with --export tsv (repeatable, combinable with json/pdf):

respro vcf \
  --project myrespro.db \
  --vcf sample.vcf \
  --ref-fasta reference.fa \
  --output my_output \
  --export tsv

Columns¶

Column	Description
`reference`	Matched internal reference name (VCF mode only; omitted in FASTA mode). Populated for multi-species VCF runs; empty for single-reference runs where the chrom is not the reference accession.
`gene`	Feature / gene name (display name applied when configured).
`nt_mut`	Nucleotide change on the internal reference, `ref{pos}alt` (1-based). Combined codon events use `ref_codon{codon_pos}alt_codon`.
`nt_mut_user`	Nucleotide change on the user-supplied reference (VCF coords before remap). VCF mode only; omitted in FASTA mode.
`strand`	Coding strand of the feature (`+`/`-`), sourced from the feature record.
`af`	Allele frequency (raw float).
`depth`	Read depth at the variant (VCF mode only; omitted in FASTA mode). Empty for combined formula rows.
`consequence`	Consequence label (`missense`, `frameshift`, …).
`aa_effects`	All amino-acid effects for the variant, as `<ref><pos><alt> \| <frequency> \| <tag>` joined with `;`. The tag is `observed` or `lower bound`. The single-exchange effect is listed first (when its amino-acid frequency is > 0), followed by every accepted combined-codon outcome. Combined formula rows join member effects with `;`.
`in_database`	`yes` when at least one rule matched (single or formula member); otherwise `no`.
`rule_type`	`single`, `formula`, `formula-member`, or `n/a` for non-hits.
`drug`	Drug name for the matched rule. `n/a` for non-hits.
`phenotype`	Rule phenotype label, stored verbatim (lowercased + whitespace-stripped) and resolved to a rank (e.g. `susceptible` / `low-level resistance` / `resistant`). `n/a` for non-hits.
`clinical_phenotype`	Clinical phenotype label (same vocabulary as `phenotype`). `n/a` for non-hits.
`ic50`	Rule IC50 value (string, may carry qualifiers). Empty for non-hits.
`fold_ic50`	Rule fold-IC50 value. Empty for non-hits.
`score`	Rule score. Empty for non-hits.
`source`	Rule source. `n/a` for non-hits.
`publications`	`\|`-joined publication identifiers (DOI, or PubMed ID as fallback). Empty for non-hits.

Row semantics¶

Single rules: one row per matching rule. A variant conferring resistance to two drugs produces two rows, each carrying that rule's own phenotype/IC50/fold-IC50/score.
Formula (combinatorial) rules: one combined row per fired formula rule. Member mutations are joined with ; in gene, nt_mut, nt_mut_user, aa_effects, af, and strand. The phenotype/IC50/fold-IC50/score come from the formula rule set (the combined call), not the individual members.
Formula-member-only variants: a variant that is only a formula member (no single rule of its own) gets a rule_type=formula-member row with in_database=yes and empty rule metric columns.
Effect-as-resistant: metadata-only synthetic hits (e.g. frameshifts classified as resistant by algorithm config) appear as rule_type=single rows with source=Metadata algorithm.

Web download

In the webapp, the TSV is produced for every profile/regenerate run alongside the HTML, PDF, and JSON artifacts. Download it from the Analyze page (single-report action bar), the Reports page under Analysis (per-row TSV link), or the batch/session "Download all" zip bundles. Run the CLI with --export tsv for command-line use.