End-to-end gene discovery
Genotypes plus field phenotypes in, annotated candidate genes out: breeding values, genomic relationships, GWAS, QTL narrowing and a plain-language biology summary in one run.
Quantitative genetics for cacao breeding
From field trials to candidate genes, on your own computer.
CacaoIQ is AI-powered software that turns raw genotype and phenotype data into corrected breeding values, GWAS and QTL maps, and annotated candidate genes, then explains what they mean. Established statistical methods, and your data never leaves your machine.
Private by design. Analyses and AI run locally.
Peer-reviewed methods. Built on open, published tools.
Validated. Reproduces published cacao results.
The problem
Breeding analysis shouldn't take a month of scripting.
A typical cacao genetics analysis means reshaping spreadsheets, stitching together R packages and desktop mapping tools, choosing models and thresholds by hand, and then searching gene databases one locus at a time.
CacaoIQ packages that whole workflow behind expert-chosen defaults, so breeders and geneticists spend their time on decisions, not file formats.
What it does
Five workflows, one application
Pick the goal. CacaoIQ looks at the data you upload and runs the right analysis for it.
Linkage map construction
SNP quality control, phasing and recombination estimation for full-sib families, with an integrated genetic map you can inspect chromosome by chromosome.
Phenotypic modeling
Spatially and design-corrected breeding values, repeatability, trait correlations, and general and specific combining ability (GCA/SCA) for factorial crosses.
Trait mapping
Already have BLUPs? Go straight to kinship-corrected GWAS with significance thresholds, Manhattan plots and candidate-gene search around every significant locus.
Diversity panel analysis
Population structure from the genomic relationship matrix, linkage-disequilibrium decay and neighbor-joining trees for germplasm collections.
A methods assistant built in
Ask what a number means in plain language. Answers come only from CacaoIQ's own methods manual, and it tells you when a question isn't covered.
See it in action
Real results, straight from the app
Witches' broom resistance in a public Colombian cacao collection: every significant locus marked on the Manhattan plot, with annotated candidate genes listed underneath.
AI that explains, not invents
Ask what a result means
The built-in methods assistant answers questions in plain language, drawing only on CacaoIQ's own methods manual, and it runs on your computer like everything else.
- Cites the manual sections each answer comes from
- Says so when a question isn't covered, instead of guessing
- AI biology summaries are drafted from the gene annotations shown beside them
How it works
Upload, confirm, read the results
- 1
Upload your files
Drag in phenotype spreadsheets and genotype files: CSV, Excel (any sheet), VCF or HapMap. CacaoIQ recognizes identifiers, traits, field positions and design factors automatically.
- 2
Confirm the setup
Review the column mapping and data-quality flags. Nothing is silently excluded or imputed. CacaoIQ recommends the model your trial design supports.
- 3
Explore the results
Interactive Manhattan plots, LOD profiles, genetic maps and ranked tables, with charts, CSVs and a written report saved for your records.
Your data stays yours
Built for proprietary breeding data
Genotypes and pedigrees are some of a breeding program's most valuable assets. CacaoIQ is a desktop application, not a cloud service.
- No uploads. Every statistical model runs on your own computer.
- Local AI. Biology summaries come from a language model running on your machine, not a cloud API.
- Fully offline. Statistics, gene annotation and AI summaries all run without an internet connection, even at a remote field station.
- No setup. R, Python and Java come bundled. No terminal, no administrator rights.
- Windows and macOS. A standard laptop with 16 GB of memory is enough.
Validation
Checked against published science
CacaoIQ's core analyses are tested against real, published cacao studies, not just synthetic data.
≈45 kb
Distance between CacaoIQ's witches' broom association hits and both chromosome 2 loci reported for a published Colombian cacao diversity panel, found independently from the public data.
r = 1.00
Agreement between CacaoIQ's combining-ability estimates (GCA and SCA) and the original commercial mixed-model analysis of a published cacao factorial trial.
See CacaoIQ on your own trial data
We'll walk your team through a live analysis and discuss how CacaoIQ fits your breeding program.
Request a demo