---
title: FROGS troubleshooting
description: >-
  Resolve common frogs pathway analysis issues including validation errors, low
  significance, and slow runs.
---

This page helps you resolve the common issues with [frogs](/docs/frogs) pathway analysis — validation errors from mismatched compound or gene identifiers, large dropped-inputs reports, low significance in the results, and slow runs.

> **Not available in chat**

> FRoGS is deprecated on the platform and its Modal backend has been taken offline — it is not registered with the orchestrator, so chat cannot call it. Existing FRoGS job results should be treated as unverified: a still-registered legacy form of the tool faked its results, returning mock predictions as `SUCCESS` regardless of input.

## Validation errors

- Ensure compound IDs match those used in target annotations and signatures.
- Use consistent gene identifiers (all HGNC or all Ensembl) across targets,
  signatures, and pathways.
- Confirm pathway sets are present and formatted correctly if you uploaded
  custom lists.

## Dropped inputs report is large

- Deduplicate compounds and signatures before upload.
- Check the report for missing targets or malformed signatures; fix and re-run.
- Tighten minimum target coverage only after validation passes cleanly.

## Empty or low-significance pathways

- Loosen p-value/FDR cutoffs slightly to confirm signal exists.
- Broaden pathway set (e.g., include GO + KEGG) to increase coverage.
- Verify signatures have sufficient completeness; raise tolerance if needed.

## Slow runs

- Reduce pathway set size or batch size for large signature matrices.
- Limit to high-confidence signatures first, then expand once parameters are
  stable.

## Export issues

- Re-run exports after changing filters or cutoffs; exports reflect the latest
  results.
- Keep compound-to-pathway contribution scores for downstream selection and wet
  lab follow-up.

 Start with default thresholds and the standard pathway set. Once
you see signal, tighten coverage and significance stepwise to avoid
over-filtering.

## Read next