by Prosilico

With the extensively validated ANDROMEDA by Prosilico software you will be able to predict, simulate and optimize the human clinical ADME/PK, exposures and doses of drug candidates, metabolites and other types of chemicals.

You get 30 ADME/PK parameters at your fingertips (see figure below), and performance and throughput superior to that of laboratory methods. You can predict batches of compounds and also include your own laboratory data.

The software is applicable for modern drugs, which often have more complex/multi-mechanistical characteristics, and works for compounds with low solubility and recovery, low CLint, significant excretion, gut-wall extraction and extensive efflux.

The German regulatory authority has approved a clinical trial application, which ADME/PK-data and -predictions were mainly based on ANDROMEDA-results.

The signature-based conformal prediction methodology produces 70 % prediction confidence intervals and guides high quality ADME/PK-optimization. The conformal prediction technology ensures that confidence intervals are adapted to the proper size according to the expected prediction errors for the compound chemical structure at hand.

ANDROMEDA by Prosilico is based on state-of-the-art machine learning and proprietary data bank with quality checked, stratified human clinical ADME/PK-data, unique algorithms and a unique integrated PBPK-model.

Customer values include:

  • Accurate predictions for modern drug candidates and drugs
  • Reduced uncertainty and risks
  • Increased data production and learning
  • Addition of missing data and sanity-checking of available lab data
  • Frontloaded and improved decision-making
  • Improved design of preclinical ADME/PK-programs
  • Savings of costs and time
  • Reduced use of animals and chemicals
  • Feeling good about saving costs, time, environment and animals and improving quality and understanding


ANDROMEDA by Prosilico is available as SaaS (cloud) and OnPrem (intranet installation) and is cost effective, and makes you feel good about saving animals.


ANDROMEDA by Prosilico is applicable at the drug design stage, lead optimization, candidate drug-selection, first-time-in-man predictions of ADME/PK, exposures and doses, and clinical data modeling & simulation, and useful in cases of missing laboratory data and for sanity-checking of available lab data. Its major prediction domain is molecular weight ca 100 to 700 g/mole, but studies with compounds with molecular weight up to 1000-1500 g/mole have also shown good results.


Prosilico has implemented measures to ensure an appropriate level of security, integrity and confidentiality. The SaaS-version is accessed via encrypted connections over HTTPS, and all data is encrypted during transport to and from the software using transport encryption (TLS). Prosilico has no access to the customer’s structures, data, results or any other confidential information of the customer. For the OnPrem-version, the customer has the responsibility to ensure adequate data security measures.

Performance and validity

In benchmarking studies, ANDROMEDA by Prosilico outperforms in vitro methodologies and animal models in the prediction of human ADME/PK-parameters. Having significantly lower prediction errors (median prediction errors are generally within 2- to 3-fold), and in addition to that, a wider scope of applicability. Scientific publications and White paper are available to support our claims.


and parameters

ANDROMEDA by Prosilico has three Modes:

1) Prediction Mode – 30 human ADME/PK parameters (including fabs, Vss, fu, Cbl/Cpl, CLint, CLH, CLR, CLbile, CL, oral F, t½, Fgut, ka, fe, extended-release potential, substrate specificities for CYPs 2C9, 2D6 and 3A4 and MDR-1, BCRP and MRP2, BBB permeability, brain binding and the unique biopharmaceutical parameter fdiss, which is the maximum fraction dissolved in the human GI tract in vivo and is distinct from conventional solubility measurements) + doses and exposures are predicted for one compound at a time. The molecular structures are either sketched in the molecular editor or inserted as SMILES (see figure below). Colored molecular elements provide guidance to chemical structural modification for optimizing ADME/PK-characteristics.

2) Prediction List Mode – 11 major primary human ADME/PK parameters are predicted for batches of 25 compounds at a time (insertion of SMILES) (see figure below).

3) Simulation Mode – 15 major human ADME/PK parameters from Prediction Mode are automatically transferred to this window and a plasma concentration-time profile is created. In this Mode the in silico estimates may be replaced and mixed with laboratory derived data. You may also explore “what-if”-scenarios with free data input.

Contact us for price proposal

Click here