Overview Data & Annotations Search & Explore Batch Annotation AI Agent Release notes

Batch Annotation

Upload a variant file and FAVOR turns it into a queryable variant list with annotations, validation metadata, and downstream analysis entry points. Use this workflow when you have more than a few variants and want one place to annotate, inspect, and analyze them.

Working offline or at WGS scale?

For files larger than the upload limit or air-gapped environments, use the FAVOR CLI — runs annotation, tissue context, and rare-variant tests on your own hardware. Existing R/STAARpipeline users can stick with the FAVOR Annotator.

Before you upload

Supported formats

CSV, TSV, VCF, and Parquet are accepted. FAVOR detects the file type and delimiter automatically.

Variant identifiers

The smoothest path is a file with rsIDs or coordinate-based variant columns that can be mapped cleanly.

Column review

If your headers are close to standard names, auto-mapping usually works. You can still override any mapping before submission.

Recommended first run

Start with a small file so you can verify the mapping and output shape before scaling up to a larger variant list.

FAVOR homepage with Batch Annotation navigation link — Open Batch Annotation from the main navigation to start a new upload.

Upload and validate

Upload the file

Drag and drop the variant list file into the upload area. FAVOR prepares a preview before you commit to the run.

Inspect the validation summary

Check the detected dataset type, row count, schema, and confidence score. This is where malformed files usually show up.

Confirm the sample values

The preview is there to help you verify that the detected column kinds match what your file actually contains.

Batch annotation upload dropzone accepting CSV, TSV, VCF, and Parquet files — Upload the variant list file. FAVOR handles common tabular and variant formats directly.

Validation summary for gwas_sumstats_ebi.tsv showing GWAS Summary Stats type, ~30 rows, 10 columns, 100% confidence, and schema preview with column kinds and samples — Review the validation summary before you proceed. It shows the inferred file type, schema, and sample values.

Map columns

FAVOR maps your incoming columns to canonical names. Review this step carefully, especially if your file uses project-specific headers or ambiguous abbreviations.

Auto-mapping

Exact and alias matches are detected automatically. This is often enough for standard GWAS or variant list files.

Manual override

If a column lands on the wrong canonical field, override it before submission. This is the last clean checkpoint before annotation runs.

Column mapping table showing original columns (CHR, POS, SNP, A1, A2, BETA, SE, P, N, INFO) mapped to canonical names (chromosome, position, rsid, effect_allele, other_allele, beta, se, p_value, n, info) with kind and source badges — Review each mapping. Auto-detected matches are shown, but you can reassign any column before submission.

Review results

When the job finishes, the result page becomes the handoff point for everything else: download the variant list, open analytics, or send it directly to the AI agent.

Analyze the variant list

Open Analytics

Use the built-in dashboard for quick summary charts, distributions, and first-pass inspection.

Stay in the browser

The analytics view is useful when you want deterministic summary views without switching into a conversational workflow.

Open in AI Agent

Use the agent when the next step is exploratory, comparative, or statistical rather than just descriptive.

In-browser analytics dashboard showing gene breakdown, consequence distribution, score distributions, and clinical significance charts — Open Analytics for a fast summary dashboard over the annotated variant list.

Analytics dashboard with charts for gene breakdown, consequence distribution, score distributions, and clinical significance — The analytics dashboard is optimized for direct inspection of distributions, counts, and top categories.

AI agent conversation analyzing the annotated variant list with natural language queries and generated visualizations — Open in AI Agent when you want follow-up questions, variant list comparisons, statistics, or graph-aware interpretation.

Jobs dashboard

The jobs dashboard is the history view for your uploads. Use it to revisit completed variant lists, filter by status, and reopen analysis later.

Start small

Try a file with a few hundred variants first. See the full workflow in seconds, then scale up. Go to Batch Annotation to upload your first file.

Engineering deep-dive

For streaming architecture, zero-allocation decoding, and horizontal scaling, see the Batch Pipeline page.