Helix StudioGenome IDE
← Docs
Helix CLI docs
Browse Helix CLI docs

CRISPR Plasmid Panel Benchmark

This benchmark ties Helix's CRISPR and prime editing outcome models to a simple synthetic plasmid library abstraction.

  • Spec template: templates/lab_crispr_plasmid_bench.helix.yml
  • Harness: benchmarks/crispr_plasmid_panel.py

Spec shape

The panel config declares one or more plasmid targets:

  • reference_sequence or reference_fasta – digital DNA window.
  • guides[] – CRISPR guides with sequence, pam_profile, and optional simulation overrides (draws, priors_profile, seed).
  • prime_pegs[] – optional prime editing pegRNAs with spacer, pbs, rtt, and optional simulation overrides.
  • Each guide/peg may carry a measurements block pointing at an amplicon counts table (counts_file, label_column, count_column, optional timepoint_h).

Running the benchmark

Example invocation:

python -m benchmarks.crispr_plasmid_panel \
  --config templates/lab_crispr_plasmid_bench.helix.yml \
  --out bench-results/plasmid_panel.json

The harness:

  • Runs helix.crispr.simulate.simulate_cut_repair for each CRISPR guide.
  • Runs helix.prime.simulator.simulate_prime_edit for each prime pegRNA.
  • Loads counts (when available), normalizes to probability distributions, and reports L1 distance and Jensen–Shannon divergence between simulated and observed outcome spectra.

The JSON output (crispr_plasmid_panel schema) records per-target/per-guide metrics plus aggregate means, suitable for CI tracking or methods figures.