MVP v0.7.34 — Promptbio v0.1 Mapper + v0.2 Diff + Issue 07 Substrate + v0.3 Evolution Loop

Selection Strategy Simulator

Compare selection strategies as an early decision engine: Monte Carlo replicates, neutral/random baselines, genomic selection mockups, OCS-like constraints, cross planning, Pareto trade-offs, risk probabilities, and minimal CRISPR-aware edit introgression.

Small populations are intentionally allowed. Try N=8 or N=12 to see rapid drift and fixation. Runs are manual: if parameters are unchanged, recalculation is skipped. The Run simulation button shows worker-pool progress. Solid lines are the current run; dotted lines show the previous run.

1. Datasynthetic, built-in real, or uploaded CSVs
2. Confirmpopulation & data summary
3. Constraintsinbreeding, diversity, risk floors
4. Runcompare strategies
5. Feasibilitywhich strategies make the cut
6. Decision Reportplain-language verdict
7. ExportJSON / summary text
8. Next stepcross plan, edit set, or new run

Simulation inputs

Founder population ? data source

Random seed ? reproducible

Population size ? 2–5000

Markers ? 10–5000

QTL count ? causal loci

Generations ? 1–200

Selection intensity ? % parents

Heritability ? 0–1

Mutation rate ? per allele

Strategy set ? core / advanced

Live histogram tracks ? AFS chart

Replicates ? 1–100

Worker count ? 0 = auto

Inbreeding limit ? risk threshold

Diversity loss limit ? risk threshold

Enable minimal CRISPR integration ?

CRISPR candidate edits ? decision-layer

Edit intro percent ? founders seeded

EU NGT regulatory context (optional — fills the regulatory section of the decision report)

Target trait class ? NGT-1 rule

Donor source ? NGT-1 rule

Variant type ? Annex I path

Patent / licensing declaration ? Issue 16

Licensing status Issue 16

Notes (optional)

Program constraints (optional — 0 = no constraint)

Hard caps from your breeding program. Strategies that fail any active constraint are marked infeasible; the decision report highlights the best feasible strategy separately.

Max inbreeding ?hard cap

Max diversity loss ?fraction of baseline

Max rare-useful loci lost ?hard cap

Min genetic gain ?floor

Min effective parents ?floor

Max combined risk ?hard cap

No run yet

Press Run simulation in the left panel to compare selection strategies on the configured population. Results will appear here.

The summary cards at the top compare best-gain / best-risk-adjusted / best-feasible recommendations.
The Decision Engine Output panel synthesises the verdict in plain language.
The Sensitivity sweep panel asks whether the recommendation holds if h² / selection / horizon / climate shifts.
The strategy table, Pareto chart, allele-frequency histogram, and per-generation charts give the visual evidence behind the verdict.
The candidate-edit table classifies each edit as EDIT / CROSS / WAIT with reason on hover (Issue 07).
Export the full run as JSON (button on the left) for follow-up analysis or audit.

If this is your first time, leave the defaults and just press Run — a balanced configuration is preloaded.

Decision engine output

Sensitivity sweep — does the recommendation hold if h² / selection / horizon shifts?

Axis Values (up to 5, comma-separated)

Axis value	Best strategy	Gain	Diversity	Inbreeding	Combined risk	Match?

Run notes

Genetic gain

trait mean vs baseline

Diversity

mean heterozygosity

Inbreeding risk

approx. diversity loss

Effective population size — Ne = 1 / (2 ΔF), log scale; dashed lines mark FAO vulnerable (Ne=100) and long-term-viability (Ne=50) thresholds

Allele-frequency drift

mean abs. shift from start

Rare useful loci lost

warning metric

Fixed loci

rapid fixation becomes obvious in tiny populations

Pareto decision chart

genetic gain vs combined risk probability — axes are pickable when multi-trait is active

Upper-left is usually better: more gain with less combined risk. White outline marks non-dominated Pareto candidates.

Live allele-frequency spectrum

updates per generation while the run is live

Waiting for first generation snapshot…

10 bins over [0, 1]: bin k counts markers whose current allele frequency falls into [k/10, (k+1)/10). Tracked from the BreedOS default ("balanced") strategy when present; otherwise the first configured strategy. The snapshot freezes on the final generation when the run completes.

CRISPR edit candidates

This is intentionally not guide design. It is a minimal decision-layer demonstration: which beneficial low-frequency loci might be worth seeding into a breeding strategy simulation.

Rank	Locus	Effect	Allele freq.	Gain score	Risk	Edit / Cross / Wait	Decision	NGT

Strategy recommendations

Strategy	Rank	Score	Final gain ±σ	Diversity ±σ	Inbreeding ±σ	Risk	P(inb/div/loss)	Fixed loci	Parents	Pareto	Feasible	Δ vs previous	Recommendation

Export & next step

The run above is now in the browser only. Export it before closing the tab or running a new configuration — there is no server-side persistence.

Next step

If the recommendation is stable across the climate sweep, plan the next breeding cycle around the best-feasible strategy named above.
If the recommendation is fragile, choose the strategy explicitly for the weather year you expect (the climate-robustness section names the alternative).
If the candidate-edit table flagged EDIT rows, route those loci to your CRISPR pipeline (Benchling / Synthego / CRISPResso) for guide-RNA design — BreedOS does not design guides.
If you uploaded predictions and the imported_gebv strategy won, the GEBVs from your prediction pipeline are doing real work in this run; document the prediction-pipeline version in the exported JSON for audit.
For a new run with different parameters: change the inputs on the left and press Run simulation again. The previous run is overlaid as dotted lines on the charts for direct comparison.