Skip to main content
skill CX Codex

tooluniverse-pharmacokinetics — Claude Code skill for PK analysis

tooluniverse-pharmacokinetics calculates pharmacokinetic parameters from measured drug concentrations over time. Use it to analyze a dose profile or estimate absolute bioavailability from IV and oral data.

Apache-2.0 140 installs ★ 1,798 updated 2026-10-24

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add mims-harvard/ToolUniverse --skill tooluniverse-pharmacokinetics -a claude-code
Open on GitHub

What tooluniverse-pharmacokinetics does

This skill runs non-compartmental analysis on concentration-time measurements. It calculates Cmax, Tmax, AUC, terminal half-life, clearance, volume of distribution, and MRT. It can also fit a one-compartment model and calculate absolute bioavailability from AUC and dose values for IV and oral administration. The analysis uses measured concentrations collected after a dose.

Send data to NCA_compute_parameters or process a CSV locally with scripts/nca_from_csv.py. Use NCA_calculate_bioavailability for the IV-versus-oral calculation. Keep time, concentration, and dose units consistent. For oral data, CL and Vd are apparent values (CL/F and Vd/F), not direct estimates. The analysis assumes a single dose; for steady state, analyze one dosing interval.

What it helps with

  • calculate Cmax, Tmax, and AUC from a concentration-time profile
  • estimate terminal half-life from the curve’s late phase
  • derive clearance and volume of distribution from IV data
  • calculate absolute bioavailability using IV and oral AUCs
  • fit a one-compartment model to concentration measurements

Example prompts for the agent

  • Calculate PK parameters from my measured concentration-time data after an IV dose.
  • Estimate absolute bioavailability from the oral and IV AUC and dose values.
  • Fit a one-compartment model to this drug concentration profile.

How to install tooluniverse-pharmacokinetics

1Skills CLI (Vercel)

npx skills add mims-harvard/ToolUniverse --skill tooluniverse-pharmacokinetics -a claude-code

2Manually: put the files in place

~/.claude/skills/tooluniverse-pharmacokinetics/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/tooluniverse-pharmacokinetics/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about tooluniverse-pharmacokinetics

How do I install tooluniverse-pharmacokinetics in Claude Code?

Run npx skills add mims-harvard/ToolUniverse --skill tooluniverse-pharmacokinetics -a claude-code. The skill file is located at ~/.claude/skills/tooluniverse-pharmacokinetics/SKILL.md.

What does tooluniverse-pharmacokinetics do?

It analyzes measured drug concentrations over time to calculate PK parameters. It also supports one-compartment fitting and absolute bioavailability calculations.

Does it report clearance from oral data?

Yes, as apparent clearance (CL/F), since oral absorption may be incomplete. Direct interpretation of clearance requires IV data.

Source text

SKILL.md · Apache-2.0
---
name: tooluniverse-pharmacokinetics
description: Pharmacokinetic (PK) analysis of concentration-time data — non-compartmental analysis (NCA) for Cmax, Tmax, AUC (0-t and 0-∞), terminal half-life, clearance (CL), volume of distribution (Vd), MRT, and absolute bioavailability (F). Also one-compartment fitting. Use when you have plasma/serum drug concentrations over time after a dose and need PK parameters, or to compute bioavailability from IV + oral AUCs. NOT for ADMET property prediction from structure (use tooluniverse-admet-prediction).
disable-model-invocation: true
---

# Pharmacokinetic (PK) Analysis — Non-Compartmental Analysis

Turn a **concentration-vs-time** profile after a dose into the standard PK parameters, and compute **bioavailability** from IV + oral data. Non-compartmental analysis (NCA) is the model-independent workhorse used for most PK reporting.

## When to use this

- You have measured plasma/serum (or other matrix) drug concentrations at known times after a dose.
- You need Cmax/Tmax/AUC/half-life/clearance/Vd, or absolute bioavailability F.
- Comparing exposure (AUC, Cmax) between formulations, doses, or routes.

This is **measured-data** PK. For predicting ADMET properties from a chemical structure, use `tooluniverse-admet-prediction`.

## Step 1 — Prepare the concentration-time data

| Issue | What to do |
|---|---|
| **Units — be consistent** | One time unit (h), one concentration unit (mg/L or ng/mL), one dose unit (mg). Pass them as `time_unit`/`conc_unit`/`dose_unit`. CL and Vd come back in derived units (e.g. L/h, L). |
| **Route matters** | Set `route` to `iv` or `po`/oral. CL and Vd are only directly interpretable for **IV** data; from oral data they are **apparent** (CL/F, Vd/F) because absorption is incomplete. |
| **Include t=0** | For IV bolus include the t=0 (back-extrapolated) point; for oral the pre-dose value is usually 0. |
| **BLQ (below limit of quantification)** | Leading BLQs before the first measurable → treat as 0; BLQs in the terminal tail → drop them (don't set to 0, it corrupts the terminal slope). |
| **Sampling design** | You need enough late points to define the terminal phase (≥3 points clearly in the log-linear decline) or the half-life and AUC0-∞ are unreliable. |
| **Single vs multiple dose** | NCA here assumes a single dose. For steady-state, analyze one dosing interval (AUC0-τ) and say so. |

## Step 2 — Run NCA

```bash
tu run NCA_compute_parameters '{
  "times":[0,0.5,1,2,4,8,12,24],
  "concentrations":[0,2.5,4.8,6.1,4.2,2.1,1.0,0.2],
  "dose":100, "route":"iv",
  "dose_unit":"mg", "conc_unit":"mg/L", "time_unit":"h"}'
```

Returns `Cmax`, `Tmax`, `Clast`, `Tlast`, `AUC0_last`, `AUC0-inf`, `AUC_extrapolation_pct`, `lambda_z`, `t_half`, `r_squared_terminal_fit`, `clearance_CL`, `volume_distribution_Vd`, `MRT_iv`, with a `units` block. AUC uses the FDA/EMA **linear-up / log-down trapezoidal** method.

For a CSV profile (with BLQ handling), `scripts/nca_from_csv.py` computes the same parameters locally.

Other tools:
- `NCA_fit_one_compartment` — fit a 1-compartment model (k, V, CL) when you want a parametric model instead of NCA.
- `NCA_calculate_bioavailability` — absolute F from `auc_po`, `dose_po`, `auc_iv`, `dose_iv` (see Step 4).

## Step 3 — Interpret the parameters

| Parameter | Meaning | Notes / sanity |
|---|---|---|
| **Cmax / Tmax** | Peak concentration & time to peak — absorption rate/extent. | For IV bolus Cmax is at t=0; a later Tmax means absorption (oral) or distribution. |
| **AUC0-t / AUC0-∞** | Total exposure (area under the curve). The key exposure metric. | AUC0-∞ extrapolates the tail using `Clast/lambda_z`. |
| **AUC_extrapolation_pct** | % of AUC0-∞ that was extrapolated beyond the last point. | **>20% → AUC0-∞ (and anything derived from it) is unreliable**; report AUC0-last instead and note insufficient sampling. |
| **lambda_z / t_half** | Terminal elimination rate constant and half-life. | Trust only if `r_squared_terminal_fit` ≥ ~0.95 and ≥3 terminal points were used. |
| **CL** (clearance) | Volume cleared per time = `Dose/AUC0-∞` (IV). | From oral data this is **CL/F** (apparent). |
| **Vd** | Volume of distribution = `CL/lambda_z` (IV). | From oral data this is **Vd/F** (apparent). |
| **MRT** | Mean residence time. | Longer MRT = slower overall elimination. |

## Step 4 — Absolute bioavailability (F)

F needs the **same drug given both IV and orally** (ideally same subjects, dose-normalized):

```bash
tu run NCA_calculate_bioavailability '{"auc_po":35.0,"dose_po":200,"auc_iv":43.4,"dose_iv":100}'
```

`F = (AUC_po / Dose_po) / (AUC_iv / Dose_iv)`. Report as a fraction or %. F near 1 = well absorbed; low F = poor absorption or high first-pass metabolism. F > 1 signals a data/dosing error (recheck units and doses).

## Step 5 — Quality gotchas (state these)

- **Extrapolation >20%** → don't report AUC0-∞/CL/Vd as reliable; the profile wasn't followed long enough.
- **Bad terminal fit** (`r_squared_terminal_fit` < 0.9, or <3 tail points) → half-life is unreliable.
- **CL/Vd from oral data are apparent** (CL/F, Vd/F) — never present them as true clearance/volume without IV data.
- **Units** drive CL/Vd — a wrong conc unit silently scales them. Always check the returned `units` block.
- **Flip-flop kinetics** (absorption slower than elimination) makes the "terminal" slope reflect absorption, not elimination — suspect it when oral t½ ≫ IV t½.

## Honest limitations

- NCA is model-independent and robust but gives no mechanistic structure (no separate absorption/distribution rate constants) — use `NCA_fit_one_compartment` or population PK for that.
- AUC accuracy depends entirely on sampling density around Cmax and in the terminal phase.
- Single-dose assumptions; for steady state analyze one interval (AUC0-τ) and accumulation separately.

## Related skills
- `tooluniverse-admet-prediction` — predict ADME properties from structure (no measured data).
- `tooluniverse-dose-response` — IC50/EC50 potency from concentration-response (not time-course).
- `tooluniverse-statistical-modeling` — compare PK parameters across groups.

© mims-harvard · Apache-2.0 · mims-harvard/ToolUniverse

Similar skills

SKILL.md skills catalog →

From the same set136

The whole set →
tooluniverse-meta-analysis tooluniverse-meta-analysis combines results from multiple studies into a pooled estimate, measures heterogeneity, and creates a forest plot. Useful for systematic reviews and replicated experiments. CX Codex +1 146installs tooluniverse-protein-structural-annotation-pdb tooluniverse-protein-structural-annotation-pdb annotates PDB protein residues by binding interface, ligand pocket presence, and solvent accessibility. CX Codex +1 145installs tooluniverse-variant-predictor-dms-validation tooluniverse-variant-predictor-dms-validation benchmarks protein variant-effect scores against experimental deep mutational scanning (DMS) records. It checks whether computational predictions match actual functional disruption. CX Codex +1 142installs tooluniverse-protein-lof-mechanism The tooluniverse-protein-lof-mechanism skill synthesizes five computational signals—from AlphaMissense through DynaMut2—to propose a specific mechanistic explanation for how a missense variant causes loss of protein function. CX Codex +1 141installs tooluniverse-protein-sae-variant-interpretation tooluniverse-protein-sae-variant-interpretation interprets protein missense variants using ESMC-6B Sparse Autoencoder features to reveal lost or gained functional biological properties. CX Codex +1 140installs tooluniverse-residue-functional-mechanism-interpretation The tooluniverse-residue-functional-mechanism-interpretation skill for Claude Code explains the significance of protein residues using structural context, UniProt annotations, and optional DMS or SAE evidence. CX Codex +1 140installs tooluniverse-mendelian-randomization The tooluniverse-mendelian-randomization skill tests whether one factor causally drives another (e.g., cholesterol → heart disease) using genetic variants as instrumental variables. CX Codex +1 137installs tooluniverse-primer-design tooluniverse-primer-design designs PCR and qPCR primers, computes Tm and Ta, and screens oligonucleotides for hairpins and dimers. CX Codex +1 131installs

Project sponsors

Companies that keep this analytics open