hcp-align solve <MODE> [INPUT] [SCORING] [OUTPUT]
Modes are edit-distance, global-linear, global-affine, local-linear, and
semiglobal-linear. Provide exactly one query source and one target source with
--query/--target or --query-file/--target-file.
Important options:
--scope alignment|score--backend auto|hcp|banded-affine--band N--cross-check--operations--integrity--format text|json|jsonl|paf|sam
Common batch options on solve, verify, and compare:
--threads N(defaultmin(available CPUs, 8))--max-in-flight N(default2 × threads)--max-in-flight-bases N(default64,000,000)--fail-fast--errors PATH(JSONL sidecar; conflicts with--fail-fast)--temp-dir PATH
Affine scoring applies gap_open + gap_extend to the first gap position and
gap_extend to each continuation.
hcp-align verify --alignment PATH --format json|jsonl|paf|sam INPUT [OPTIONS]
--level optimal is the default. --max-cells defaults to 25,000,000 and 0
means unlimited. PAF and SAM additionally require --mode and applicable
scoring flags. JSON v2 records embed their problem definitions.
Imported records resolve query and target sequences by ID. Duplicate or missing IDs are errors. PAF/SAM reverse-strand records are rejected in v0.2 rather than being silently mis-normalized.
JSON and JSONL are framed incrementally. JSONL, PAF, and SAM have recoverable
line boundaries; JSON-array syntax corruption is fatal. JSONL continues after
record failures by default and emits ordered v2 error records inline. Other
output formats require --errors to continue.
hcp-align compare \
--left LEFT --left-format FORMAT \
--right RIGHT --right-format FORMAT \
INPUT
Both records are exactly verified. Different traces with the same valid optimum are reported as tie-breaking differences and remain equivalent. JSON comparison artifacts embed normalized sequence text and omit timings and expanded operation arrays so they can be attached to an issue and replayed without the original FASTA/FASTQ files.
| Code | Meaning |
|---|---|
| 0 | Requested assurance passed for every record. |
| 1 | Invalid, incorrectly scored, or non-optimal record. |
| 2 | Any error record, or a usage, parsing, source-resolution, or configuration error. |
| 3 | Inconclusive verification or internal verification error. |
Precedence is 2, then 1, then 3, then 0.