Drug Discovery · openadmet
Octant_CYP_inhibition_reactivity_blog_release
Cytochrome P450 inhibition and metabolic reactivity screening data across diverse small molecules for ADMET modelling.
What Octant_CYP_inhibition_reactivity_blog_release is
Cytochrome P450 (CYP) enzymes mediate the oxidative metabolism of most therapeutic drugs and represent a major source of adverse drug-drug interactions. Despite the necessity of predicting CYP interactions during drug discovery, public datasets often suffer from sparsity, assay noise, and heterogeneous measurement protocols. This dataset, developed by Octant Bio for the OpenADMET consortium, provides self-consistent experimental measurements across approximately 1,200 diverse small molecules tested within a single high-throughput platform.
The resource captures quantitative CYP3A4 inhibition alongside reaction phenotyping for both CYP3A4 and CYP2J2, an enzyme central to extra-hepatic metabolism that remains underrepresented in public screening data. To support both machine learning model training and assay reproducibility research, the data are organized into compound-level summary metrics alongside full well-level experimental readouts, outlier flags, and quality control annotations.
How it was made
Experimental screens were conducted in miniaturised 1536-well microplate formats using Gentest Supersomes. CYP3A4 inhibition was assessed in 4 µL reaction volumes through 12-point fluorescence-based dose-response assays, incorporating a 30-minute pre-incubation with active enzyme to measure both reversible and time-dependent inhibition mechanisms. Reaction phenotyping was conducted in 2 µL volumes, quantifying parent compound depletion through contactless Echo acoustic ejection mass spectrometry (Echo-MS) with four biological replicates per condition.
The dataset is structured across five subsets. Compound summaries include 'inhibition' (1,340 fitted pIC50 curves) and 'reactivity' (2,442 compound-enzyme substrate depletion pairs). Corresponding well-level subsets ('inhibition_detailed' with 16,931 raw fluorescence rows and 'reactivity_detailed' with 19,344 Echo-MS peak area rows) provide raw optical signals, plate spatial coordinates, and outlier designations. An auxiliary subset, 'will_it_fly' (11,353 rows), details MS ionisation efficiency across two buffer carriers (ammonium fluoride and ammonium formate) to evaluate compound detectability prior to enzymatic assays.
Key facts
- Target enzymes
- CYP3A4 and CYP2J2
- Assay modalities
- 12-point fluorescence dose-response curves and Echo acoustic ejection mass spectrometry (Echo-MS)
- Assay format
- 1536-well microplates (2 µL to 4 µL reaction volumes)
- Compound count
- Approximately 1,200 diverse small molecules
- Data format
- Tabular CSV
- Identifier
- DOI: 10.57967/hf/9647
What you can use it for
- Predictive CYP3A4 inhibition modelling
- Train quantitative structure-activity relationship (QSAR) and machine learning models on fitted pIC50 values, Hill slopes, and maximum effect values derived from standardized compound SMILES.
- Substrate phenotyping and clearance prediction
- Identify small molecule substrates and predict metabolic depletion percentages against hepatic CYP3A4 and extra-hepatic CYP2J2.
- Assay quality control and spatial plate artifact analysis
- Utilise raw well-level fluorescence and mass spectrometry replicates to study edge effects, evaluate custom outlier-filtering routines, and model experimental noise.
- Mass spectrometry ionisability benchmarking
- Assess chemical detectability and ionisation propensity across distinct carrier solvents using the 11,353 pre-screened compound records in the 'will_it_fly' subset.
Explore the data
The schema, real rows, full-text search and a SQL filter — read in your browser, nothing downloaded.
Explore
Read the real rows without downloading anything
Reading rows…
LiveRead from openadmet/Octant_CYP_inhibition_reactivity_blog_release at the moment you asked. Nothing is cached or stored — every row above came from that request.
How the columns are distributed
Over 1,340 rows of inhibition/train
CYP3A4_pIC50
float- min
- 1.80
- median
- 5.36
- mean
- 5.23
- max
- 7.47
19.1% empty
CYP3A4_pIC50_ci_lower
float- min
- 1.03
- median
- 5.32
- mean
- 5.10
- max
- 7.43
19.1% empty
CYP3A4_pIC50_ci_upper
float- min
- 2.99
- median
- 5.40
- mean
- 5.35
- max
- 7.51
19.1% empty
CYP3A4_pIC50_se
float- min
- 9.7e-3
- median
- 0.02
- mean
- 0.07
- max
- 0.68
19.1% empty
activity_status
string_label- YES89%
- NO11%
direction
string_label- DOWN89%
- FLAT11%
drc_qc_flag
string_label- PASS85%
- FAIL INCOMPLETE14%
- FAIL BASELINE1%
drc_qc_status
string_label- PASS85%
- FAIL15%
emax_log2fc
float- min
- -11.34
- median
- -3.70
- mean
- -3.80
- max
- -1.01
ocnt_batch
string_text- min
- 19
- median
- 19
- mean
- 19
- max
- 19
plate_qc_status
string_label- PASS93%
- FAIL7%
qc_flag_primary
string_label- PASS85%
- INCOMPLETE14%
- BASELINE1%
rollover_status
string_label- NO99%
- YES1%
saturation_status
string_label- NO58%
- YES41%
- UNKNOWN1%
slope_log2
float- min
- -0.49
- median
- 0.46
- mean
- 0.49
- max
- 2.60
standardized_smiles
string_text- min
- 19
- median
- 56
- mean
- 55.01
- max
- 173
Strengths and limits
Good at
- Generated on a single experimental platform using consistent reagent batches and protocols, eliminating batch disparities common to aggregated public datasets.
- Captures combined reversible and time-dependent inhibition through an active-enzyme 30-minute pre-incubation step.
- Provides paired compound-level summary endpoints and granular well-level raw signals for transparent quality control.
- Includes reaction phenotyping for CYP2J2, expanding metabolic data for an enzyme poorly represented in public chemical repositories.
- Supplies solvent ionization benchmarking across over 11,000 molecules to clarify mass spectrometry compatibility.
Watch out for
- Inhibition IC50 metrics combine reversible and time-dependent inhibitory mechanisms, making them non-equivalent to conventional reversible-only IC50 measurements.
- Screening is limited to two cytochrome P450 isoforms (CYP3A4 and CYP2J2), omitting other major drug-metabolising isoforms such as CYP2D6, CYP2C9, or CYP1A2.
- Compounds that failed to ionise above background in Echo-MS were excluded from reactivity reporting rather than flagged as zero, restricting the measured chemical space to mass-spectrometry-detectable species.
Questions
Can I look inside Octant_CYP_inhibition_reactivity_blog_release without downloading it?
Yes. This page reads its schema, real rows and per-column statistics in the browser, with full-text search and a SQL filter. Nothing is downloaded to your machine.
What licence is Octant_CYP_inhibition_reactivity_blog_release under?
Its authors declare cc-by-4.0. Read the terms on the original page before reusing the data.
Why do the CYP3A4 IC50 values differ from standard reversible inhibition values?
The assay incorporates a 30-minute pre-incubation step with active enzyme before measuring substrate turnover, causing the resulting IC50 to capture both reversible inhibition and time-dependent inhibition.
What is the purpose of the 'will_it_fly' subset?
It assesses the mass spectrometry ionisation efficiency of roughly 11,000 library compounds across two carrier buffers (ammonium fluoride versus ammonium formate) to identify molecules that cannot be reliably detected before running enzymatic depletion assays.
What enzymes are evaluated in the reactivity phenotyping assays?
Reactivity is measured against CYP3A4 and CYP2J2 via acoustic ejection mass spectrometry to quantify compound depletion relative to no-enzyme controls.
Can models be trained directly on raw experimental readouts instead of fitted summaries?
Yes, the 'inhibition_detailed' and 'reactivity_detailed' subsets provide unaggregated well-level fluorescence intensities and Echo-MS peak areas alongside well coordinates and outlier flags.
How to cite it
OpenADMET and Octant Bio, 'Building the OpenADMET Data Engine', OpenADMET Blog, 2026. DOI: 10.57967/hf/9647.
Written from Octant_CYP_inhibition_reactivity_blog_release's own dataset card. Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.