CacaoIQ™by Scilytics

Methods, validation and citations.

The method at each stage, the published studies CacaoIQ was checked against, and the data, genome and software to cite when you publish results.

Methods by stage

The methods manual that ships with the app describes each step. How Scilytics chooses and checks its methods is on the Science page.

StageMethodSoftware
Data checks and BLUPsLinear and generalized mixed models with row, column and block terms; ordinal models for scored traitslme4, ordinal
Genomic relationshipsVanRaden genomic relationship matrix with SNP quality controlNumPy
Genomic heritabilityGenomic REML (GREML) on the genomic relationship matrix by default; a Bayesian animal model for binary (liability scale), count and repeated-measure traits, for confirmed multi-tree plots, or when you choose itNumPy/SciPy (REML), MCMCglmm
Association mappingMixed-model GWAS (EMMAX) with kinship correction and a Bonferroni threshold; genomic-control λ reported as a check on inflationNumPy/SciPy (EMMAX)
Linkage mappingPhase and recombination fractions for full-sib (CP) families, markers in fixed physical order, Kosambi distancesLep-MAP3 (OneMap data format for fullsibQTL)
QTL mappingInterval mapping for outcrossed full-sib families with a genome-wide permutation threshold (permutations by Haley-Knott regression), LOD-support intervals, variance explainedfullsibQTL
Combining abilityFactorial GCA/SCA mixed model with confidence intervals for each parent's GCA and each family's SCA; narrow-sense h² and dominance ratio d²lme4
Diversity analysisRelationship-matrix PCA, LD decay, neighbor-joining treesNumPy, Biopython
Candidate genesGene models from the Matina 1-6 v1.1 cacao reference, with UniProt and InterPro functional annotations bundled for offline useBundled data
Gene summaryOne line per gene near the peak, quoted from the bundled UniProt and InterPro annotation. A short domain note, labeled as reviewed, appears only for domains in a reviewed glossaryBundled data
Methods assistantAnswers questions about the methods from the user manual, citing the section, and suggests column roles when a file is loaded; runs on your computerPhi-4-mini, run locally with llama.cpp

Published studies, rerun.

CacaoIQ was run on the public data of these studies and its results compared with the published ones.

Validation on public data. Sources are listed under References below.
StudyWhat was comparedPublishedCacaoIQ
Osorio-Guarín et al. 2020GWAS, cacaoMarker-trait associations; 102 accessions, 8,131 SNPs10 associations10 of 10: the same SNP, in CacaoIQ’s top 12
Bekele et al. 2022Trait correlations, cacaoPublished Pearson correlations between traits6 correlations6 of 6, to 3 decimals
John & Williams 1995α-lattice trial, oatGenetic variance, against the reference value distributed with the data0.142902 reference value0.14290
Baek et al. 2026Repeatability, cacaoPublished repeatability across four harvests, from CacaoIQ’s input with the paper’s formula4 values4 of 4 reproduced

Some workflows have no public cacao data yet. We list only the comparisons we could check, and we name the crop for each one.

References

The data, genome and software CacaoIQ builds on. The method papers are on the Science page.

Validation data

  1. Osorio-Guarín JA, et al. (2020). Genome-wide association study reveals novel candidate genes associated with productivity and disease resistance to Moniliophthora spp. in cacao (Theobroma cacao L.). G3: Genes, Genomes, Genetics 10(5): 1713-1725. doi:10.1534/g3.120.401153
  2. Bekele FL, Bidaisee GG, Allegre M, et al. (2022). Genome-wide association studies and genomic selection assays made in a large sample of cacao (Theobroma cacao L.) germplasm reveal significant marker-trait associations and good predictive value for improving yield potential. PLOS ONE 17(10): e0260907. doi:10.1371/journal.pone.0260907
  3. Baek I, Upadhyay RK, Kim MS, et al. (2026). Longitudinal genomic prediction of cacao disease resilience identifies robust witches’ broom disease targets. Frontiers in Plant Science 17: 1837102. doi:10.3389/fpls.2026.1837102
  4. John JA, Williams ER (1995). Cyclic and Computer Generated Designs, 2nd ed. Chapman & Hall. Oat α-lattice data as distributed in the R package agridat (john.alpha). CRAN.R-project.org/package=agridat

Reference genome and annotation

  1. Motamayor JC, Mockaitis K, Schmutz J, et al. (2013). The genome sequence of the most widely cultivated cacao type and its use to identify candidate genes regulating pod color. Genome Biology 14: r53. doi:10.1186/gb-2013-14-6-r53
  2. UniProt Consortium. UniProt: the Universal Protein Knowledgebase. www.uniprot.org
  3. InterPro: protein families, domains and functional sites. EMBL-EBI. www.ebi.ac.uk/interpro/

Software

  1. Harris CR, et al. (2020). Array programming with NumPy. Nature 585: 357-362. doi:10.1038/s41586-020-2649-2
  2. Virtanen P, et al. (2020). SciPy 1.0: fundamental algorithms for scientific computing in Python. Nature Methods 17: 261-272. doi:10.1038/s41592-019-0686-2
  3. Cock PJA, et al. (2009). Biopython: freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics 25(11): 1422-1423. doi:10.1093/bioinformatics/btp163
  4. Bates D, Mächler M, Bolker B, Walker S (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67(1): 1-48. doi:10.18637/jss.v067.i01
  5. Christensen RHB. ordinal: Regression models for ordinal data. R package. CRAN.R-project.org/package=ordinal
  6. Hadfield JD (2010). MCMC methods for multi-response generalized linear mixed models: the MCMCglmm R package. Journal of Statistical Software 33(2): 1-22. doi:10.18637/jss.v033.i02
  7. Rastas P (2017). Lep-MAP3: robust linkage mapping even for low-coverage whole genome sequencing data. Bioinformatics 33(23): 3726-3732. doi:10.1093/bioinformatics/btx494
  8. Margarido GRA, Souza AP, Garcia AAF (2007). OneMap: software for genetic mapping in outcrossing species. Hereditas 144: 78-79. doi:10.1111/j.2007.0018-0661.02000.x
  9. Gazaffi R, Margarido GRA, Pastina MM, Mollinari M, Garcia AAF (2014). A model for quantitative trait loci mapping, linkage phase, and segregation pattern estimation for a full-sib progeny. Tree Genetics & Genomes 10: 791-801. doi:10.1007/s11295-013-0664-2
  10. llama.cpp. github.com/ggml-org/llama.cpp
  11. Abouelenin A, et al. (Microsoft) (2025). Phi-4-Mini Technical Report. arXiv:2503.01743. arxiv.org/abs/2503.01743

Data in the screenshots

The screenshots on this site use public data from these sources:

  • Data: Osorio-Guarín J.A., Berdugo-Cely J.A., Coronado-Silva R.A., Baez E., Jaimes Y., Yockteng R. (2020). Supplemental Material for Osorio-Guarín, Berdugo-Cely et al., 2020. GSA Journals, doi:10.25387/g3.10028999.v1, licensed under CC BY 4.0. Article: G3 10(5): 1713-1725, doi:10.1534/g3.120.401153. Analyzed with CacaoIQ.
  • Data: Bekele F.L., Bidaisee G.G., Allegre M., Argout X., Fouet O., Boccara M., Saravanakumar D., Bekele I., Lanaud C. (2022). S10 Table, Genotype data used in GWAS, licensed under CC BY 4.0. Supporting information to: Genome-wide association studies and genomic selection assays made in a large sample of cacao (Theobroma cacao L.) germplasm reveal significant marker-trait associations and good predictive value for improving yield potential. PLOS ONE 17(10): e0260907, doi:10.1371/journal.pone.0260907. Analyzed with CacaoIQ.

Questions about the methods or software: contact@scilytics.ai