Input data formatting

Input Data Format

Both the CLI and Python API work with the same underlying table structure.
The CLI reads from TSV/CSV files, while the API can take a pandas.DataFrame directly.

Every table must include:

1. Gene column
   Contains the gene identifier for each row.
   The column name can be anything; it is specified when running the tool.

2. Feature column
   Contains the feature identifier (for example, isoform or transcript ID).
   The column name is also user-specified.

3. Expression data
   One or more numeric columns containing either raw counts or normalized TPM values.
   The type (counts vs TPM) is specified at runtime.

---

Single-Sample Format

For datasets representing one sample, include exactly one numeric expression column.

Example:

gene_id    transcript_id    sample1
G1         T1               100
G1         T2                50
G2         T3                30
G2         T4                 0

Multi-Sample Format

For datasets with multiple samples, include one column per sample, and all must be numeric.

Example:

gene_id    transcript_id    sample1    sample2    sample3
G1         T1               100        80         90
G1         T2                50        20         30
G2         T3                10         5          8
G2         T4                 0         2          4

Notes

Column names for gene, feature, and samples are flexible — you specify them when using either the CLI or the API.

All expression columns must contain only non-negative, numeric values.

For API usage, simply pass a properly structured pandas.DataFrame; no file format or separator concerns apply.