Pauling vs. Atomwise

Pauling puts the full screening pipeline in your hands: screen millions of compounds yourself, iterate in real time, and validate hits with MD and ADMET without waiting on anyone. Atomwise pioneered deep-learning virtual screening (AtomNet) with strong pharma partnerships, but it's a service -- you send them targets, they send back hits.

Category Pauling Atomwise
Access modelSelf-serve platform -- run jobs on demandService model -- submit targets, wait for results
Screening methodPhysics-based docking (UniDock/Vina) with explicit scoring functionsDeep learning (AtomNet CNN) trained on co-crystal structures
Screening scale10M+ compounds via Cloud Dataflow auto-scalingLarge library (proprietary), but you don't control scope
Turnaround timeMinutes to hours depending on library sizeWeeks to months per engagement
Iteration speedModify parameters and re-run immediatelyNew round requires new service request
MD validationFull GROMACS MD with MM-PBSA binding energyNot included -- docking scores only
ADMET profilingIntegrated ADMET pipeline on every hitNot part of standard deliverable
Pose quality controlPoseBusters + MolProbity validation on all poses
Data ownershipFull ownership -- private catalogs, all files yoursAtomwise retains rights to screen data
TransparencyEvery parameter, input, and output logged in jobs dashboardBlack-box predictions -- no scoring function visibility
Hit rate enrichmentExplicit Vina scoring with known physicsStrong enrichment from AtomNet in benchmarks
Pharma partnershipsEarly-stage -- focused on computational chemistsExtensive pharma collaborations (Bayer, Merck, etc.)