gut_wall_cleaned / README.md
jiosephlee's picture
Upload folder using huggingface_hub
f5a0ce1 verified
|
Raw
History Blame Contribute Delete
1.55 kB
---
pretty_name: Gut Wall Cleaned
tags:
- chemistry
- pharmacokinetics
- tabular
---
# Gut Wall Cleaned
This dataset contains 1,490 accepted finite scalar measurement children from the
versioned canonical-base pipeline. A parent row with several unambiguous measurements
can produce several child rows. Rejected, malformed, ambiguous, bounded, range-only,
TDC-overlapping, and endpoint-unassignable records are not published here.
## Cleaning
The following source columns are replaced in place with validated canonical or lexical
normalizations:
- `smiles``canonical_smiles`
- `gut_wall_process``canonical_endpoint_key`
- `transporter_or_enzyme``transporter_or_enzyme_normalized`
- `substrate_status``substrate_status_normalized`
- `assay_system``assay_system_normalized`
- `intestinal_site``intestinal_site_normalized`
- `qualifying_conditions``qualifying_conditions_normalized`
- `measured_value``scalar_value`
The public column names on the left are retained; the names on the right identify the
validated internal values used to replace them. Columns omitted from this dataset:
- `extraction_id`: `internal_identifier`
- `global_identifier`: `internal_identifier`
- `auc_window`: `all_null_in_dataset`
- `defining_timepoint`: `all_null_in_dataset`
`canonical_endpoint_key` is the endpoint identity field. No `condition_key` is added.
`species_exact` is populated only for explicit, unambiguous species references. Narrative
and other source fields without a validated normalization remain source text.