Skip to content

ROADMAP.md

Latest commit

 

History

History
80 lines (60 loc) · 3.3 KB

ROADMAP.md

File metadata and controls

80 lines (60 loc) · 3.3 KB

Roadmap

This document describes the planned development of sharkmer across upcoming major releases. See the issue tracker for detailed specifications.

v2.0 — Cleanup and polish (released)

Released. See CHANGELOG.md for details.

v3.0 — Graph traversal (released)

Released. See CHANGELOG.md for details.

Key improvements: bidirectional graph extension, frontier-queue extension eliminating O(n²) node scanning, reachability pruning replacing ad-hoc heuristics, coverage-aware tip clipping, read threading with bubble resolution, composite path scoring, mismatch-aware primer kmer cap, dynamic node budget, read caching, and panel versioning and validation infrastructure. New c_elegans panel.

Paired-end phasing (#101) is deferred to v4.0: the infrastructure (PairedEndLink, thread_reads_paired()) is built but downstream consumption is not yet wired.

v4.0 — Metagenomics

The goal of v4.0 is to reliably produce multiple distinct products from mixed samples.

Target improvements:

  • Paired-end phasing (#101): wire PairedEndLink data into bubble resolution and path scoring (infrastructure built in v3.0)
  • Read-supported graph pruning (#99): prune edges/nodes lacking read support to reduce chimeric joins
  • Deconvolution of mixed-organism samples based on coverage profiles
  • Support for reporting multiple distinct amplicons per gene
  • Per-product coverage profiles to distinguish true variants from assembly artifacts
  • Configurable limits for number of amplicons per gene
  • Validation against known metabarcoding datasets
  • Document QIIME2/SILVA workflow and consider convenience flags (#68)

v5.0 — New targeting paradigms

The goal of v5.0 is to generalize beyond the PCR metaphor. The paired-primer sPCR workflow remains a primary feature, but v5.0 adds additional targeting modes built on the underlying seeded assembly engine.

Oligo-based extensions

  • Single-primer (forward-only) seeded assembly — relax the requirement for a reverse primer, extending outward from a single seed until coverage drops or a length limit is reached
  • Synthetic ddRADseq — seed with restriction enzyme recognition sites to produce in silico RAD-like fragments
  • Synthetic targeted enrichment — seed with UCE (ultraconserved element) oligo probe sequences to recover UCE loci and their flanking regions from genome skimming data

Exotic targeting

Novel approaches to isolating homologous genome regions across samples that have no parallel in benchtop methods but are natural extensions of seeded assembly:

  • Seed generation from reference sequences, alignments, or gene models (may require helper scripts or a companion tool)
  • Cross-sample kmer comparison to identify conserved seeds without a reference
  • Iterative seeding where products from one round seed the next

Architecture considerations

  • The targeting mode (paired-primer, single-oligo, restriction site, UCE probe) should be a parameter of the panel/primer YAML format introduced in v2.0, not a separate code path — the seeded assembly engine is shared
  • v3.0 graph traversal improvements and v4.0 multi-product support are prerequisites — single-primer extension and UCE enrichment both produce variable-length products that need robust graph handling