Input Data Formats¶

This guide provides detailed specifications for preparing input data for tfbpmodeling analysis.

Overview¶

tfbpmodeling requires two main input files plus optional supplementary files:

Response File: Gene expression data (dependent variable)
Predictors File: Transcription factor binding data (independent variables)
Blacklist File (optional): Features to exclude from analysis

Response File Format¶

Structure¶

The response file contains gene expression measurements:

gene_id,pTF1
YBR001C,0.234
YBR002W,-0.456
YBR003W,1.234
YBR004C,-0.123

Requirements¶

Element	Requirement	Description
Format	CSV	Standard comma-separated values
`gene_id` column	Gene identifiers	Must match predictor file exactly
Response column	Named for the perturbed TF (e.g. `pTF1`)	Must match `--perturbed_tf`; numeric expression values

Predictors File Format¶

Structure¶

The predictors file contains transcription factor binding data:

gene_id,TF_1,TF_2,TF_3,TF_4,pTF1
YBR001C,0.123,0.456,0.789,0.012,0.345
YBR002W,0.234,0.567,0.890,0.123,0.456
YBR003W,0.345,0.678,0.901,0.234,0.567
YBR004C,0.456,0.789,0.012,0.345,0.678

Requirements¶

Element	Requirement	Description
Format	CSV with comma separators	Standard comma-separated values
First Column	Gene identifiers	Must match response file exactly
Header Row	TF names	Transcription factor identifiers
Data Cells	Numeric values	Binding scores
No Missing Values	Complete data required	All cells must contain numeric values

Gene Identifier Consistency¶

Critical Requirements¶

Both files must use identical gene identifiers:

# Verify gene ID consistency
response_df = pd.read_csv('response.csv', index_col=0)
predictor_df = pd.read_csv('predictors.csv', index_col=0)

# Check for exact matches
common_genes = set(response_df.index) & set(predictor_df.index)
response_only = set(response_df.index) - set(predictor_df.index)
predictor_only = set(predictor_df.index) - set(response_df.index)

print(f"Common genes: {len(common_genes)}")
print(f"Response only: {len(response_only)}")
print(f"Predictor only: {len(predictor_only)}")

Blacklist File Format¶

Structure¶

Simple text file with one gene identifier per line:

YBR999W
YCR888X
ribosomal_protein_L1
housekeeping_gene_1
batch_effect_gene