pytdc

Therapeutics Data Commons. AI-ready drug discovery datasets (ADME, toxicity, DTI), benchmarks, scaffold splits, molecular oracles, for therapeutic ML and pharmacological prediction.

1.16x

Quality

52%

Does it follow best practices?

Impact

91%

1.16x

Average score across 3 eval scenarios

Securityby

Advisory

Suggest reviewing before use

Optimize this skill with Tessl

npx tessl skill review --optimize ./scientific-skills/pytdc/SKILL.md

Evaluation results

100%

30%

Drug-Target Interaction Model Evaluation

DTI cold split evaluation

Criteria

Without context

With context

Multi-pred import

100%

Cold drug split

100%

Drug_ID column access

100%

Target_ID column access

100%

TDC Evaluator used

100%

Evaluator named metric

100%

Zero drug overlap verified

100%

get_split seed param

100%

results.json produced

100%

Dataset name BindingDB_Kd

100%

82%

-2%

ADMET Property Benchmarking Pipeline

ADMET benchmark 5-seed protocol

Criteria

Without context

With context

admet_group import

100%

admet_group path param

100%

Exactly 5 seeds

100%

Predictions keyed by seed

33%

16%

group.evaluate() called

46%

group.get() for dataset

100%

benchmark_results.json produced

100%

Mean and std extracted

100%

No manual metric computation

100%

91%

Candidate Molecule Screening with Property Oracles

Multi-objective oracle screening

Criteria

Without context

With context

Oracle import

100%

Target-binding oracle used

100%

QED oracle used

100%

SA oracle used

100%

Batch oracle evaluation

100%

SA score inverted

50%

Multi-objective composite score

100%

screening_results.json produced

100%

Repository: K-Dense-AI/claude-scientific-skills
Commit: b58ad7e

Evaluated: 27 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Drug-Target Interaction Model Evaluation ADMET Property Benchmarking Pipeline Candidate Molecule Screening with Property Oracles

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.