The drug discovery software market divides into three categories: computational chemistry suites for structure-based design, AI-native platforms for generative chemistry and target identification, and R&D data management systems for ELN, LIMS, synthesis tracking, data analysis, and compound registration.
We compared the top 8 drug discovery platforms across features, pricing, and deployment models.
Top 8 drug discovery software deployment and pricing comparison
Product | Deployment | Free/Academic Tier | Starting price | Use cases |
|---|---|---|---|---|
BIOVIA Discovery Studio | Cloud/On-premise | Free visualizer | N/A | End-to-end simulation platform spanning target ID to lead optimization with built-in ELN via 3DEXPERIENCE |
ChemAxon | Cloud/On-premise | No | N/A | Cheminformatics toolkit with Marvin editor and JChem registration |
Cresset Flare | Desktop | No | N/A | Electrostatic field-based ligand design platform with XED force field |
Dotmatics | Cloud | No | N/A | Scientific informatics platform with ELN, bioregistry, and assay data management |
OpenEye Orion | Cloud-only | No | N/A | Cloud-native Orion platform with shape-based screening and OEChem toolkit |
Recursion OS | Cloud/Internal | No | N/A | Phenomics platform with high-content imaging and BioHive supercomputing |
Schrödinger Suite | Desktop/On-premise | No | $7,500/year (30 tokens) | Physics-based computational chemistry suite featuring FEP+ for binding affinity prediction |
StarDrop | Desktop/Cloud | No | $10,000/year/user | Multi-parameter optimization platform with Glowing Molecule visualization |
Note: The products are listed alphabetically.
Drug discovery software feature comparison
BIOVIA Discovery Studio
BIOVIA Discovery Studio offers an end-to-end pipeline from target identification through lead optimization. The tool integrates with the Dassault Systèmes 3DEXPERIENCE platform for enterprise data management and regulatory compliance.
BIOVIA Discovery Studio covers the main parts of computational drug discovery:
- Simulations: Molecular dynamics, free energy calculations, and other simulation tools for studying molecular behavior and interactions.
- Structure-based design: Tools for protein-ligand docking, fragment-based design, and optimizing compounds using 3D structural information.
- Ligand- and pharmacophore-based design: Methods for de novo drug design, activity profiling, multi-target design, and finding molecules with desired interaction patterns.
- Biotherapeutics and antibody modeling: In silico tools for antibody modeling, protein engineering, and biologics optimization.
- Macromolecule design and analysis: Tools for analyzing and designing proteins, nucleic acids, and other large biological molecules.
- QSAR, ADMET, and predictive toxicology: Predictive insights for pharmacokinetics, safety, toxicity, and drug-like properties.
- Visualization: A free molecular design visualizer for viewing, manipulating, and analyzing biological and chemical structures.
BIOVIA Discovery Studio is best suited for large enterprises that require regulatory compliance.
Figure 1: BIOVIA Discovery Studio simulations dashboard.1
ChemAxon
ChemAxon provides cheminformatics infrastructure, including its Marvin chemical structure editor and JChem engines, for chemical data standardization. The platform supports cloud and on-premise deployment with Java and REST APIs for integration.
Chemaxon’s Marvin real-life example:
A large global pharmaceutical company used Chemaxon’s Marvin chemical editor to improve chemical data management inside a desktop data visualization and analytics application.
The company required a chemical drawing tool that could integrate smoothly with its existing infrastructure before an upcoming go-live deadline. The requirements included support for SMILES/SMARTS notation, reaction mapping, stereochemistry handling, Markush structure enumeration, and a strong .NET API for integration.
Chemaxon implemented Marvin for more than 300 users and upgraded a smaller group of Marvin JS users to the newer Marvin environment. The company integrated the .NET API add-on into the company’s existing system, and the full integration with the desktop analytics application was completed in about one to two weeks.
The project helped the company meet its launch timeline without compromising functional, technical, or business requirements. It also simplified licensing by consolidating Chemaxon licenses into a single 19-month term, reducing procurement complexity.2
Chemaxon is best for organizations needing chemical data standardization and API integration.
Cresset Flare
Cresset Flare’s electrostatic field-based ligand modeling using the XED force field enables scaffold hopping and ligand design without reliance on protein crystal structures. Flare supports RBFE calculations and macrocycle conformer generation within its FEP framework.
- 2D Interaction Maps: Summarizes ligand-protein interactions in a clear 2D view for easier interpretation.
- Activity Atlas: Provides qualitative SAR insights to help understand how molecular changes affect activity.
- Activity Miner: Identifies activity and selectivity cliffs in SAR datasets.
- FieldTemplater: Helps predict binding modes when crystallographic protein structure data is unavailable.
- Free Energy Perturbation (FEP): Supports lead optimization by predicting which ligand changes are most likely to improve binding.
- AI coding assistant: Supports scripting, workflow automation, and method customization by helping users write or refine code for Flare-related analysis workflows.
Cresset Flare is best suited for medicinal chemists working without crystal structures.
Figure 2: Flare AI coding assistant example.3
Dotmatics
Dotmatics provides an integrated scientific informatics platform that spans ELN, BioRegister, compound registration, and assay data management, with Studies and Vortex visualization. The platform offers integration across the Dotmatics portfolio, including Geneious and Prism. Dotmatics serves large biopharma companies, CROs, and industrial R&D organizations requiring a governed system of record.
Dotmatics Luma is an AI-native, multimodal R&D platform that integrates scientific data, workflows, analytics, and AI tools into a single environment. It is designed to help research teams move from raw lab data to usable scientific insight faster. Luma works in four main steps:
- Data capture: Luma connects to lab instruments, ELNs, registries, CRO uploads, files, scientific applications, and external systems. Luma Lab Connect can collect data from file-based instruments, Windows or Linux folders, S3 buckets, APIs, and SQL/JSON/CSV sources.
- Data processing: Once data is captured, Luma parses raw files, extracts metadata, and converts instrument outputs into structured, analyzable formats.
- Data harmonization and management: Luma brings together different data types, including structured, semi-structured, unstructured, sequence, numeric, text, image, and metadata.
- Analysis and AI-enabled insights: Researchers can search, visualize, query, model, and analyze harmonized data within the platform or via APIs. Luma also supports natural-language querying and generative AI use cases, helping scientists explore complex relationships across datasets.
Dotmatics is best for large biopharma companies and CROs that require a governed system of record.
OpenEye Scientific Suite Orion
OpenEye Scientific, now part of Cadence Molecular Sciences, differentiates through its cloud-native Orion platform and developer-focused OEChem toolkit. The suite includes ROCS for shape-based screening, EON for electrostatic comparison, OMEGA for conformer generation, and FRED for docking.
Orion operates on AWS and Cadence OnCloud with no on-premise installation option, targeting organizations building custom computational pipelines.
OpenEye Scientific Suite Orion is best suited for developers building custom pipelines.
Figure 3: Orion’s 3D simulation and analysis dashboard.4
Recursion OS
Recursion OS enables a massive phenotypic dataset generated by high-throughput automated imaging of cellular phenotypes, processed via computer vision and the BioHive supercomputing infrastructure. The platform includes the Map of Biology, which visualizes biological relationships, and has generated approximately 65 petabytes of proprietary data.
Recursion LOWE:
LOWE is Recursion’s LLM-orchestrated Workflow Engine, an AI-enabled system within the Recursion OS platform designed to support complex drug discovery workflows through natural-language interaction.
It enables researchers to query Recursion’s biological and chemical datasets, explore potential drug-target relationships, generate and prioritize novel compounds, assess properties such as ADMET and solubility, and coordinate downstream activities, including synthesis planning and experimental execution.
LOWE functions as an intelligent workflow orchestration layer that connects Recursion’s proprietary datasets, predictive models, generative chemistry capabilities, and laboratory operations.5
Recursion is best for rare disease and drug repurposing programs.
Schrödinger small-molecule drug discovery suite
Schrödinger differentiates itself through physics-based Free Energy Perturbation (FEP+) calculations that predict binding affinity. The suite integrates Glide for docking, WaterMap for hydration thermodynamics, and Prime for protein structure prediction within the Maestro graphical interface.
Schrödinger’s proprietary program real-life example:
Schrödinger’s digital chemistry platform helped identify SGR-1505, a novel MALT1 inhibitor, as a development candidate in 10 months. The program focused on MALT1, a target implicated in lymphocyte regulation and relevant to relapsed or refractory B-cell malignancies, including chronic lymphocytic leukemia. Earlier MALT1 inhibitor approaches faced issues with drug-like properties, so the goal was to find a potent small molecule with a better balance of potency, permeability, solubility, and overall developability.
The team used a design-predict-make-test-analyze workflow supported by physics-based modeling, machine learning, predictive ADMET models, and data analytics. They computationally evaluated more than 8 billion compounds, used WaterMap to analyze the binding site, applied de novo design and synthetically aware enumeration to generate ideas, and used FEP+ to predict relative binding affinity. LiveDesign was used to centralize modeled and experimental data for collaborative decision making.
In the first three months, the team evaluated more than 1,700 molecules using Active Learning FEP+ and identified two novel potent MALT1 inhibitor series after synthesizing fewer than 50 compounds. After that, they used multiparameter optimization to balance potency, solubility, and permeability. The team assessed more than 5,000 ideas, and 43 compounds met the program criteria, and only a smaller subset moved into synthesis and testing.
The outcome was SGR-1505, selected within 10 months after 78 compounds were synthesized in the lead series and 129 compounds across the full program. Schrödinger presents the case as evidence that combining large-scale computational screening, physics-based prediction, machine learning, and collaborative informatics can reduce the number of compounds that need to be synthesized while accelerating the path from hit discovery to a development candidate.6
Schrödinger is best suited for pharma and biotech teams requiring high-accuracy potency modeling.
Optibrium StarDrop
Optibrium StarDrop specializes in multi-parameter optimization (MPO) for lead optimization. The platform offers both desktop and cloud deployment with modular pricing for ADMET, generative chemistry, and 3D design modules.
adMare with StarDrop real-life example:
adMare’s, a Canadian life science company, work spans early lead identification through to clinical candidate selection, requiring chemists to evaluate compound potency, ADME characteristics, physicochemical properties, selectivity, and broader structure-activity relationships. StarDrop supports this process by helping researchers organize, visualize, and interpret complex compound datasets more efficiently.
A notable application is patent analysis. When chemists extract large numbers of compounds from patent literature, StarDrop’s clustering, similarity analysis, chemical space visualization, and Card View features help identify relevant starting points and understand how compound series have been optimized.
The team also uses StarDrop to examine SAR trends, compare pIC50 values, predict properties such as logP and logD using ADME QSAR, prepare compound libraries for docking studies, and screen virtual libraries with eSim3D.7
Optibrium StarDrop is best suited for medicinal chemists prioritizing ADMET and lead optimization.
Regulatory and compliance considerations for AI-assisted drug discovery
Regulatory agencies have begun formalizing guidance for AI/ML in drug development. In January 2025, the FDA published draft guidance on “Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making,” proposing a risk-based credibility assessment framework for AI models used in nonclinical, clinical, and manufacturing contexts.8 The guidance explicitly excludes drug discovery activities, focusing only on data supporting regulatory decisions.9
In January 2026, the FDA and EMA jointly published “Guiding Principles of Good AI Practice in Drug Development,” establishing ten high-level principles covering human-centric design and proportional validation requirements across the medicines lifecycle.10 The agencies emphasized that AI systems should support, not replace, human judgment, with validation requirements scaled to the AI system’s potential impact.11
Liens de référence
Soyez le premier à commenter
Votre adresse courriel ne sera pas publiée. Tous les champs sont obligatoires.