Integrated statistical genetics · Human, animal & plant

From genetic association to genomic prediction.

Run reproducible workflows for GWAS, polygenic scores, TWAS, fine-mapping, Mendelian randomization, and genomic selection across human, animal, and plant research.

Explore analyses

6 analysis families 3 species contexts 1 reproducible job record

Human, animal, and plant genetics connected through a shared DNA analysis workflow — Harmonized inputs Modular analysis Interpretable results

Analysis workspace

Choose a starting point

Each workflow includes input validation, method configuration, job monitoring, and reproducible results.

Read the quick manual

One evidence graph for complex traits

From variant signals to biological decisions.

Build, compare, and connect polygenic scores, transcriptome-wide associations, credible sets, causal estimates, and genomic predictions—without rebuilding the workflow for every method.

Design an analysis Explore the platform

9analysis families

3species contexts

1reproducible manifest

Modular workflow human · plant · animal

Summary statisticsGWAS · eQTL · QTL

Individual datagenotype · phenotype

Multi-omicsexpression · annotation

Reference panelsancestry · breed · line

PRSsingle · multi-trait · cross-ancestry

TWASgenes · tissues · omics

Fine-mappingcredible sets · colocalization

MRcausal effects · sensitivity

Genomic selectionBLUP · Bayes · ML · DL

Evidence score0.87

risk prediction prioritized genes credible variants causal pathways

✓ Frozen models & complete provenance

PPolygenic scoresC+T · stacking · ancestry transfer TTWAS & omicsPrediXcan · FUSION · TIGAR FFine-mappingGIFT · SuSiE · FINEMAP · coloc MMendelian randomizationIVW · Egger · sensitivity LGWAS & LDSCQC · heritability · genetic correlation GGenomic selectionDMU · PBLUP · GBLUP · SSBLUP

Harmonize once

Detect formats, align alleles and genome builds, validate IDs, and register reusable LD and omics resources.

Compose analyses

Mix prediction, association, fine-mapping, causal inference, and multi-omics engines in a single project manifest.

Interpret together

Compare methods, connect variant-to-gene-to-trait evidence, and export dashboards, models, logs, and provenance.

Analysis workspace

OmniGenet analysis

Configure input data, select statistical-genetics methods, review the generated job, and submit it to the OmniGenet server.

OmniGenet analysis

PRS · Single-trait

Upload and validate summary statistics, then select methods and run the analysis.

Email optional

Inputs and species registry

Species reference Project manifest path/file Genotype / marker path/file Phenotype and covariate path/file GWAS/QTL/omics summary path/file Expression/protein/methylation path/file Pedigree / relationship path/file LD panel / annotation path/file

Upload input file

Files are saved under the backend resources folder and can be used by the paths below.

File type Choose file

No file uploaded yet.

Evidence-based upload diagnostics After upload, OmniGenet will check file structure, missingness, identifiers, and supported genetics-specific fields.

Method engines

Genomic prediction · Single-trait Compare statistical, machine-learning, and deep-learning prediction engines for one target trait.

Upload GWAS

Upload summary statistics directly. OmniGenet detects the format, checks scientific fields, and inserts the stored path into the selected workflow.

Upload once for plots, PRS, TWAS, fine-mapping and MR

Trait Dataset role Choose File Download sample Separator

Column identifiers are suggested after inspection. Review the mapping before building the analysis plan.

No summary-statistics file uploaded.

Summary-statistics readiness check After upload, OmniGenet will inspect identifiers, effect columns, allele fields, missingness, and basic format consistency.

Method readiness Upload the required summary-statistics files to see which selected methods can run now and which need an LD panel, expression model, or additional dataset.

Fields marked with an asterisk (*) are required; all others are optional.

* rsID identifier * P-value identifier * Beta / OR identifier * Non-effect allele * Effect allele Chromosome identifier Position identifier * Sample size * Ancestry Genome build

GS / genomic selection

Configure breeding-value and genomic-prediction engines for local, server-side, or HPC runs.

Analysis engine

DMU

Run genetic evaluation through blupADC::run_DMU. Defaults use the official phenotype, pedigree, and macOS/Linux/Windows DMU software provided by blupSUP.

blupADC · DMU analysis_model phe_path phe_name phe_col_names target_trait_name fixed_effect_name random_effect_name covariate_effect_name genetic_effect_name relationship_path relationship_name DMU module DMU software path dmu_algorithm_code integer_n missing_value iteration_criteria Output directory

Machine Learning

Machine-learning genomic prediction with regularized linear, kernel, bagging, and boosting models.

LightGBM SVR Elastic Net Random Forest GBM XGBoost Phenotype file Marker file Target trait ID column Test size Random seed Marker preprocessing Elastic Net alpha / l1_ratio RF trees / mtry Boosting parameters SVR parameters Cross-validation Feature importance Output path / prefix

PRS

Choose polygenic-risk-score engines and scoring strategies for human, animal, or plant prediction studies.

Summary-first mode: target genotype and phenotype are optional. With summary statistics alone, OmniGenet produces downloadable SNP-weight files. Upload a target cohort only when individual scoring or accuracy evaluation is required.

Single-trait PRS parameters

One discovery GWAS is sufficient for SNP-weight generation. Standard PRSice-2 uses one optional target cohort; an independent tuning split is available under advanced validation.

Discovery summary stats Complementary GWAS Trait labels Ancestry labels Target genotype prefix (optional) Target phenotype file (optional) Phenotype column Outcome type LD reference panel DBSLMM executable DBSLMM PLINK 1.9 DBSLMM LD-block directory DBSLMM heritability h² DBSLMM settings mtPGS executable PLINK 1.9 executable mtPGS LD reference prefix LD block file Genetic covariance matrix Environmental covariance matrix Target trait Trait sample sizes mtPGS settings P-value thresholds Clumping settings Execution mode PRSice.R path (optional) PRSice executable path Score normalization Output path / prefix

Fine mapping

Specify credible-set, cross-population, and colocalization engines for locus-level interpretation.

Fine-mapping engines

Select a fine-mapping workflow to see its compatible methods.

Wakefield ABF credible set built-in GIFT conditional TWAS integrated R SuSiE SuSiEx MESuSiE FINEMAP mvSuSiE flashfm Colocalization ABF built-in

Fine-mapping parameters Trait name Locus name Summary stats file Secondary GWAS Population / ancestry Sample size N Locus selection Auto locus window (kb) Chromosome Region start Region end LD matrix file Locus list file Region definition Credible set level Effect prior variance Max causal variants LD matrix source Coloc priors Report threshold Output path / prefix

Choose methods and run

Analysis summary

Review the selected methods, resolved inputs, expected figures, and output location. OmniGenet validates all requirements automatically before analysis.

Backend GS / ML-DL

Selected method

Trait GY1

Output gs_wheat_lightgbm

Resolved inputs Awaiting uploaded data

Automatic output Figures, tables, and an HTML report

Advanced · Reproducibility filesManifest, command, and local/HPC runner Expand

TSV task rows remain available inside the downloaded bundle and are submitted to the backend automatically.

Server execution

OmniGenet automatically validates the workflow first. Analysis starts immediately after all preflight checks pass.

Advanced options

Validation only—do not run analysis or create figures

Queued Results

Job ID

Running Status:

Enter a Job ID to inspect status and outputs.

Job Basic Information

Job ID—

Analysis—

Created—

Execution—

Status—

Return code—

Output

No job selected.

Workflow execution

Submit, monitor, and reproduce each analysis

OmniGenet creates a job manifest, validates inputs, launches the selected method engines, records logs, and keeps the generated runner script for reproducible local or HPC execution.

Define the workflow

Select datasets, QC modules, method engines, integration modules, and requested result products.

Run through the server

The backend writes job-specific configuration, starts a dry-run, streams logs, and can launch the real analysis.

Export reproducibly

Download model lists, manifests, logs, result files, and runner scripts for local reruns or auditing.

bash run_omnigenet_job.sh
RUN_REAL=1 bash run_omnigenet_job.sh

Post-analysis dashboard

Explore and export completed results

OmniGenet returns interpretable result summaries, downloadable machine-readable files, and full run metadata for each submitted job.

Integrated evidence score 0.87

Evidence by analysis

Fine-map0.91

TWAS0.83

MR0.76

PRS0.68

Prioritized evidence

Target	Evidence	Type
GENE1	0.94	TWAS
rs1248	0.89	PIP
PATHWAY7	0.81	MR

Download package metrics.json predictions.tsv feature_importance.tsv run_manifest.yaml

Method engines

One workspace for prediction, association, discovery, and interpretation

Genetic evaluation

Run pedigree, genomic, and single-step evaluation workflows such as PBLUP, GBLUP, SSBLUP, DMU, BLUPF90, and GCTA.

Genomic selection

Compare breeding-value prediction models across traits, populations, breeds, lines, or environments.

Machine and deep learning

Train LightGBM, SVR, MLP, DNNGP, and future model families with shared preprocessing, validation, and reporting.

Association and discovery

Connect GWAS/QTL, TWAS, fine mapping, Mendelian randomization, and marker or gene prioritization workflows.

Omics integration

Combine marker, phenotype, expression, protein, methylation, annotation, and pathway evidence into ranked candidates.

Benchmarking and reports

Summarize accuracy, stability, runtime, feature importance, diagnostics, downloadable tables, and reproducible logs.

Help and resources

Templates, tutorials, API access, and reproducible execution

1Download a harmonized GWAS summary-statistics template for plotting, PRS, TWAS, fine-mapping, and MR. 2Download the complete GWAS plots, PRS, TWAS, fine-mapping, colocalization, and MR demo-data bundle.

3Job status pages with progress states, logs, warnings, and downloadable run manifests.

4API endpoints for upload, job submission, status polling, result listing, and result download.

5Read the summary-statistics guide: required inputs, built-in methods, external resources, outputs, and caveats.

6Reproducible runner scripts for local machines, servers, and HPC environments.

Local and server execution

Run the same analysis through the web or command line

The included FastAPI backend accepts uploads, builds job-specific manifests, runs validation and Snakemake dry-runs, streams logs, and exposes result files for download.

Rscript Scripts/run_blupadc_dmu.R --config config.yaml --dry-run
Rscript Scripts/run_blupadc_dmu.R --config config.yaml

python3 Scripts/run_ml_gs.py --config config.yaml --dry-run
python3 Scripts/run_ml_gs.py --config config.yaml

Research assistant

OmniGenet AI

Explain Use configured remote model Response

Checking AI availability…

OmniGenet AI

I can retrieve reviewed OmniGenet knowledge and explain upload QC, genomic selection/blupADC, PRS, TWAS, fine-mapping, MR, job logs, and structured results.

Open diagnostics workspace

Only redacted structured summaries are sent. Raw summary-statistics rows are not included. Research support only—not clinical diagnosis.

From genetic association to genomic prediction.

Choose a starting point

From variant signals to biological decisions.

Harmonize once

Compose analyses

Interpret together

OmniGenet analysis

PRS · Single-trait

Inputs and species registry

Upload input file

Universal QC and harmonization

Method engines

Upload GWAS

Phenotype summary

GS / genomic selection

PRS

GWAS visualization

TWAS / omics

LDSC

Fine mapping

MR

Result integration

Outputs and reports

Choose methods and run

Analysis summary

Server execution

Job Basic Information

Output

Submit, monitor, and reproduce each analysis

Define the workflow

Run through the server

Export reproducibly

Explore and export completed results

One workspace for prediction, association, discovery, and interpretation

Genetic evaluation

Genomic selection

Machine and deep learning

Association and discovery

Omics integration

Benchmarking and reports

Templates, tutorials, API access, and reproducible execution

Run the same analysis through the web or command line