What We Do

We develop potential therapeutics through our extensive researcher and foundation networks.

We identify research that can be readily translated to treatments for rare and neglected diseases.

We provide the drug discovery preclinical and clinical expertise to move therapeutics towards the clinic.

We can help you by providing our expertise for your grant submission.

How We Report

Understand what the customer wants

Understand the problem and not just what our models can do

Clearly describe what we did:

  • What did we find?
  • How does this benefit the company?
  • How can we build on this work for the company?

Provide graphical representations of the data

Highlight the advantages and acknowledge limitations

What We Have Delivered

We have obtained 9 orphan drug designations.

We have delivered measurable results for our collaborators and clients.

We have widely published our results in respected peer reviewed journals.

AI Safety

Collaborations Pharmaceuticals, Inc. has gained considerable expertise regarding AI safety which has led to our involvement in developing “Recommendations for generative AI: Preventing AI from creating chemical threats”:

Learn from The Hague Ethical Guidelines

Engage numerous AI ethics institutes or other experts to provide guidance

Increase ethical training for computing students and raise awareness

Increase training of scientists in companies to recognize potential for dual use of generative AI

Keep a human in the loop

Waitlist restriction (e.g. like GPT-3 was initially) to limit access

Use a public facing API to control access and how models are used

Federated learning - use encrypted data to train model without decrypting data

Disclosure of potential for dual use in publications to encourage recognition of this potential and visibility

Regulation of software and applications in industry/academia: limit access to tools, knowledge and expertise 

Urbina, F., Lentzos, F., Invernizzi, C., Ekins, S. Dual use of artificial-intelligence-powered drug discovery. Nat Mach Intell 4, 189–191 (2022).

Urbina, F., Lentzos, F., Invernizzi, C., Ekins, S. A teachable moment for dual-use. Nat Mach Intell 4, 607 (2022).

Urbina, F., Lentzos, F., Invernizzi, C., Ekins, S. AI in Drug Discovery: A Wake-up Call. Drug Disc Today, 28, 103410 (2023).

Urbina, F., Lentzos, F., Invernizzi, C., Ekins, S. Preventing AI from Creating Biochemical Threats. J Chem Inf Model, 63, 691-694 (2023).

Ekins, S., Lentzos, F., Brackmann, M., Invernizzi, C. There’s a ‘ChatGPT’ for biology. What could go wrong? Bulletin of the Atomic Scientists (2023).

Ekins, S., Brackmann, M., Invernizzi, C., Lentzos, F., Generative Artificial Intelligence-Assisted Protein Design Must Consider Repurposing Potential GEN Biotechnology, 2, 296-300 (2023).

Ekins S., Biosecurity and Artificial Intelligence in the Life Sciences GEN Biotechnology, 3. 284-289 (2024). 

Ekins S., The dark side; dual use implications of generative drug discovery, 163-174 in Ekins S., An Introduction to generative Drug Discovery, CRC press, Boca Raton FL., 2025. 

Client Use Cases

Company needed a commercial read across tool integrated with machine learning models for regulatory filings. We developed a new tool that used the EPA software and data in a secure environment. We enable the company to predict how their molecules are likely to behave when the EPA does read across so they can anticipate their questions.

Chemicals

Client needed new linkers for PROTACS with ideal predicted properties. We developed a generative design approach to design linkers and predict solubility. Provided new ideas, IP to client. There is no commercial tool to do what we provided.

Drug Discovery

Curated data on endocrine disruption (ER, AR, Aromatase) built models and predicted for their 1000s of ingredients. Enabled them to prioritize ingredients for in vitro testing. Saved them significant time and $ in testing compounds - also reduced animal usage. Published papers, talks at conferences, showed thought leadership for client.

Consumer Products

Company needed a way to score 1bn molecules for toxicity properties. We developed an approach to screen 1bn molecules in a few hours. We also developed a toxicity screen with 42 properties. Generated a massive amount of data that enabled the company to raise a VC $30M A round

DNA-encoded Library Company

CONTRACT MACHINE LEARNING

Collaborate to Accelerate Drug Discovery

  1. We sign a confidentiality agreement
  2. We send you a quote for the work
  3. You send us the molecules of interest
  4. We send you the predictions and applicability domain data etc.

Interested?

CONTACT US