Projects

Things I've
built.

Projects that started as a question at the bench and turned into something useful. More on the way.

01 // Molecular docking

DuganDock

A desktop molecular docking workbench. Describe an experiment in plain English — "dock DHT, testosterone and estradiol at the androgen receptor" — and it finds a suitable structure in the PDB, resolves every compound against PubChem and ChEMBL, and hands back a plan to review before anything runs. Then it drives the whole pipeline unattended: structure prep, pocket finding, protonation and stereoisomer enumeration, docking, rescoring with a second function, pose physics checks, and a self-contained report.

The part I care about most is that it tells you when not to trust it. Every run automatically redocks the crystal ligand as a control, scores that pose against the real geometry, and grades the run on what its own evidence can actually support — so a result that looks clean but rests on three compounds says so out loud.

Plain English inDescribe the experiment; get a reviewable plan card with structures, molecules, rationale and caveats.
Nothing inventedEvery structure comes from a database and is checked against a provenance ledger — no chemistry written from memory.
Built-in controlThe crystal ligand is redocked every run, with a symmetry-aware RMSD and a pass/fail verdict.
Honest about limitsA trust grade from geometry, scoring, physics, agreement, resolution and sample size — plus what the run can't answer.
Second opinionVina and Vinardo over the same poses, with optional Boltz-2 on the GPU when a run earns it.
ResumableEvery ligand is a row with a status. Kill it, restart it, it picks up where it left off.
AutoDock VinaVinardoBoltz-2RDKitPDBFixerPDB · PubChem · ChEMBLSQLitePython
DuganDock plan card for an androgen receptor campaign DuganDock run log showing pipeline stages DuganDock results table with docking scores DuganDock score distribution and ligand efficiency charts DuganDock run overview with provenance
Ask — the plan, with its caveats, before anything runs
02 // Pharmacokinetics

Serum

A pharmacokinetics modeling tool for research peptides. Enter a dosing schedule and Serum plots the predicted concentration curve over time — a one-compartment model with first-order absorption and elimination, each administration contributing independently and summed by superposition. The dashboard tracks which compounds remain above zero, when the next administration falls, and how close a compound is to steady state, alongside any body metrics recorded over the same window.

The regimen simulator is the piece I built it for: model up to four hypothetical schedules side by side and read off the steady-state behavior — Cmax, Cmin, Cavg, fluctuation, accumulation ratio, and time to 90% of plateau — as a purely theoretical comparison, separate from any recorded data. It ships with a reference library of published half-life, time-to-peak and bioavailability values, each tagged with a confidence level, plus 3D crystal structures pulled from the PDB. Everything stays in the browser: no account, no server, no tracking.

Superposition modelC(t) = Σ F·D · (ka/(ka−ke)) · (e−ket − e−kat), summed across every dose.
Regimen simulatorCompare up to four theoretical schedules and read their steady-state metrics side by side.
Reference libraryPublished PK parameters with confidence ratings, editable, plus your own entries.
Real structuresCrystal structures fetched from the RCSB PDB and rendered in 3D.
Honest about the modelPopulation-average parameters in relative units — an educational modeling tool, not a medical device or a substitute for assayed bloodwork.
Local onlyAll records in localStorage with JSON and CSV export. No account, no server, no tracking.
One-compartment PKLinear superpositionSteady-state metricsRCSB PDBJavaScriptlocalStorage

Built for experimental and educational modeling of published pharmacokinetic parameters. Not a medical device, and not medical advice.

Serum dashboard with modeled concentration curve and recorded metrics Serum regimen simulator comparing two dosing schedules Serum compound library with pharmacokinetic parameters Serum body metrics plotted against the modeled timeline Serum 3D crystal structure viewer
Dashboard — modeled levels and recorded metrics on one timeline
More

Physical builds too

Open the 3D gallery →