---
title: Inputs and data prep
description: >-
  Prepare compound-target annotations, gene signatures, and pathway sets for
  frogs pathway analysis.
---

You provide compound-target annotations, gene expression signatures, and the pathway sets you want scored. Keeping identifiers consistent across targets, signatures, and pathways is what most affects the result — mismatched identifier types are the main cause of dropped rows and partial analyses.

> **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.

## Webapp input methods

The .bloom* webapp provides a comprehensive interface for frogs pathway analysis.

## Webapp parameters

### Required file inputs

| Parameter | Description | Helper Text |
|-----------|-------------|-------------|
| **Compound List File** | File path listing compounds for cross-validation | `--cpdlist_file` |
| **Target File** | File path for compound target annotations | `--target_file` |
| **Signature File** | L1000 gene signatures file path | `--sig_file` |
| **GO-trained Gene Embeddings** | Path to GO-trained gene embeddings | `--emb_go` |
| **ARCHS4-trained Gene Embeddings** | Path to ARCHS4-trained gene embeddings | `--emb_archs4` |
| **Output Directory** | Directory to save predicted target ranking lists | `--outdir` |
| **Model Directory** | Directory to save trained models | `--modeldir` |

### Perturbagen type selection

| Option | Description |
|--------|-------------|
| **shRNA** | Short hairpin RNA knockdown (default) |
| **cDNA** | Complementary DNA overexpression |

### Training parameters

| Parameter | Range | Default | Description |
|-----------|-------|---------|-------------|
| **Number of Epochs** | 1+ | 100 | Training iterations |
| **Number of Processes** | 1-16 | 1 | Parallel processes (`--nproc`) |

### Logging options

- **Enable detailed logging to file**: Toggle for verbose output

### Personalized settings

- **Use Personalized Settings**: Toggle for project-specific configuration from Supabase

## Submit button

- **Label**: "Start FRoGS Analysis"
- **Disabled when**: Any required field (compound list, target, signature, embeddings, directories) is empty

## Required inputs

- **Compound list**: IDs used throughout your workspace; include any aliases if
  multiple systems are used.
- **Target annotations**: Mappings from compounds to targets (genes/proteins).
  Ensure consistent identifier types (e.g., HGNC symbols).
- **Gene signatures**: Expression signatures or embeddings aligned to the same
  identifiers; typically GO/ARCHS4-style vectors or ranked gene lists.
- **Pathway set**: Choose from provided sets (e.g., GO/KEGG) or upload a custom
  list of pathways/terms with member genes.

## Optional inputs

- **Weights per compound**: If some compounds are controls or should be
  down-weighted.
- **Pathway filters**: Include/exclude lists to focus on relevant biology.
- **Metadata**: Cohort labels or experiment IDs to keep outputs traceable.

## Formatting tips

- Keep identifiers consistent across targets, signatures, and pathways (e.g.,
  all HGNC or all Ensembl).
- Remove duplicates before upload; frogs will drop exact duplicates but warns on
  conflicts.
- For ranked gene lists, include directionality if available; for embeddings,
  confirm dimension compatibility with the workspace.

 frogs validates inputs and reports any dropped rows (missing
targets, malformed signatures). Fix reported issues and re-run to avoid partial
analyses.

## Read next