Molecular Design

AI molecular design for drug candidates

Blackthorn AI designs, scores and ranks small molecules, peptides, protein binders and aptamers. We deliver a shortlist of candidate structures or sequences, the evidence behind the ranking and the information your lab or CRO needs to plan synthesis and testing.

A fixed-fee feasibility check takes 2–3 weeks and starts from €5,000 / £4,700 / $6,000.

  • 2–3 weeksFeasibility check
  • 4Molecule types
  • RankedSynthesis shortlist

When should you bring in a molecular design team?

Bring us in when you have a target or an existing lead and need to decide what to make next. We can propose new candidates, design variants around a lead or narrow a large set to the compounds and sequences your lab can realistically test.

  • 01

    You have a target and need a starting point

    Define the desired properties and generate candidates against those criteria.

  • 02

    Your lead has a specific weakness

    Design variants around a property you need to improve, such as predicted stability or selectivity.

  • 03

    Your assay budget limits the first batch

    Prioritise candidates against shared criteria and keep backup options where the evidence supports them.

  • 04

    Your next design round depends on assay results

    Use the measurements to decide which candidates to keep, revise or drop.

In brief

What do you get from an AI drug design round?

You get a ranked shortlist of structures or sequences for synthesis, the scores and rationale behind each recommendation, and a handoff with the specifications your lab or CRO needs. The report records trade-offs, uncertain predictions and the questions the next experiment should answer.

  • Candidate designs

    Identifiable structures or sequences, with design origin and relevant predicted properties.

  • A ranked shortlist

    Primary candidates, suitable backups and a clear rationale for the proposed test order.

  • Lab handoff

    Candidate specifications, assumptions and open questions for the experimental team.

Is a candidate design round right for your team?

A design round fits when your team can define what the molecule should do and has a credible route to test it. You can start with a target, a reference scaffold, a candidate set or results from an earlier screen. The check establishes what those inputs can support.

You need this if

  • You need new structures or sequences, or a new set of variants around a lead.
  • You can agree on the properties that matter and the trade-offs you will accept.
  • Your lab or CRO can synthesise and test the candidates.

Skip it if

  • Your immediate need is to establish an assay or collect the first experimental evidence.
  • Your main need is a wider discovery programme or R&D software. See Drug Discovery & Development.

Delivery path

How do we go from a target to a synthesis shortlist?

We agree on the target, design criteria and experimental constraints during feasibility. Then we generate candidates, filter and rank them, and package the shortlist for your lab or CRO. Once measurements are available, a follow-on round can use those results to design and rank new variants.

  1. 01

    Check feasibility

    2–3 weeks · fixed fee

    Review the target, reference candidates, available data and practical constraints. Assess what can be compared or designed with the evidence you have.

    You get a written proceed or pause recommendation, a proposed design brief and a quote for the next round.

  2. 02

    Design candidates

    Quoted from the check

    Generate new structures or sequences, or develop variants of an existing scaffold. Track where each design came from.

    You get a candidate set tied to your design criteria.

  3. 03

    Filter and rank

    Within the quoted round

    Remove candidates outside the agreed constraints. Rank the rest on relevant predicted properties and practical synthesis requirements.

    You get a shortlist with scores, rationale, trade-offs and backups where warranted.

  4. 04

    Hand off for testing

    Within the quoted round

    Review the candidate files and key uncertainties with your experimental team. Agree which candidates to make and what the first tests should answer.

    Your lab or CRO gets candidate specifications and testing priorities.

  5. 05

    Redesign from results

    Quoted follow-on round

    Use the assay results your team shares to revisit the original ranking and decide which changes to explore.

    You get revised designs and a ranked shortlist for the next test round.

Which types of molecules can you design?

We design small molecules, peptides, protein binders and aptamers. The choice depends on the target, the intended use and what your lab can test. We can start from a brief for a new candidate, design variants of a known lead or rank an existing set before synthesis.

Small-molecule design
Propose structures or analogues and prioritise them against the agreed property profile and synthesis constraints.
Peptide engineering
Design new peptide sequences or variants and assess the properties that matter for the planned tests.
Protein and binder design
Design new protein sequences or variants and use predicted structures to help select candidates for testing.
Aptamer design
Design nucleic acid sequences from a reference scaffold or from scratch and rank them for the intended use.
Candidate screening
Compare an existing panel against shared criteria to choose which candidates should go into the next assay.
Lead redesign
Use assay measurements and prior design history to propose and rank the next set of variants.

What your lab receives

We agree on the files and specifications with your team before the design round.

  • Candidate filesStructures for small molecules or sequences for peptides, proteins and aptamers.
  • Ranking rationaleRelevant predicted properties, trade-offs and reasons to keep candidates in reserve.
  • Handoff notesSynthesis constraints, what needs experimental confirmation and how to interpret the scores.

What should your molecular design brief include?

Tell us what you want the candidate to do, what is already known about the target and how your lab will test it. Share the reference structures or sequences and assay results you can provide. Include practical limits such as the number of designs the first batch can accommodate.

Target and intended function
The biological question, target information and what a useful molecule would do.
Starting material
Reference structures, sequences, scaffolds or hits where available.
Experimental evidence
Existing assay results, including failed designs or negative controls if available.
Design constraints
Desired properties, tolerable trade-offs and synthesis or production limits.
Test plan
Who will run the first experiments, what the assays measure and what would justify another round.

How much does AI molecular design cost?

The fixed-fee feasibility check starts from €5,000 / £4,700 / $6,000 and takes 2–3 weeks. We use the check to quote the first design round, including the shortlist and lab handoff. We quote later rounds from the work they require, with monthly capacity available in story points.

  • Feasibility check

    From €5,000, fixed fee

    Review the target, inputs and test constraints. Receive a design brief, a written proceed or pause recommendation and a quote for the next round.

    2–3 weeks

  • Candidate design round

    Quoted from the check

    Generate or refine structures or sequences, then deliver a ranked shortlist, scoring rationale and synthesis handoff.

    Scope, delivery date and outputs agreed upfront

  • Lab-informed redesign

    Quoted from the check or prior round

    Review the assay data, design variants and prioritise the next synthesis batch. Monthly capacity is available for recurring rounds.

    160 story points per monthly workstream

We quote in EUR, GBP or USD. Budget separately for synthesis and lab testing.

Story points size the agreed design work. One workstream provides 160 story points of capacity each month.

What has Blackthorn AI delivered on design projects?

For a biotech client, we delivered a ranked set of aptamer designs and explained what needed to be checked in the lab. A separate small-molecule project used transcriptomic evidence to generate and prioritise compounds. Its published case study reports laboratory-confirmed activity for some of the candidates.

Molecular design project results
ProjectReported outcomeSource
Aptamer design for a biotech clientRanked candidate sequences with a rationale for the synthesis shortlist and open questions for testing.Anonymised engagement
Small molecules for skin researchGenerated and prioritised small molecules. The case study reports assay activity for some of the candidates.Read case study
Candidate generation from multi-omics dataGenerated and scored drug-like molecules and prepared candidates for lab testing.Read case study

What did candidate selection look like in practice?

An aptamer design brief asked whether candidates could fit the intended assay as well as bind the target. That constraint shaped which sequences we put forward for synthesis. In a separate small-molecule programme, transcriptomic and pathway data guided which compounds were generated and prioritised.

  • AptamersAnonymised engagement
    Biotech client

    Ranking sequences for the intended assay

    We designed candidate sequences and ranked them for the planned use. In the handoff, we explained the ranking and flagged predictions that still needed experimental confirmation.

  • Small moleculesPublished case
    Skin rejuvenation research

    Using biological evidence to select compounds

    We used transcriptomic and pathway data to guide small-molecule generation and selection. The published case reports activity in lab tests. Read case study →

Should you commission designs, use a platform or hire?

Commission a design round when you want a specialist team to deliver candidates for a defined target and test plan. A platform fits when your scientists can use it to design and evaluate candidates themselves. An in-house hire suits teams that expect a sustained volume of computational design work.

Commissioned molecular design, platform and in-house hire compared
CriterionCommission a design roundUse a platformHire in-house
Best fitA defined target and a lab decision to make.A design task covered by the tool.A continuing need across programmes.
Who designs and ranks?Blackthorn AI works to criteria agreed with your scientists.Your team runs the tool and interprets its outputs.Your staff designs candidates and owns the ranking.
What do you get?Structures or sequences, a ranked shortlist and a lab handoff.A licence and the features included in the plan.An internal capability to develop over time.
How do you start?A fixed-fee feasibility check, then a quote for the design round.Evaluate coverage and available support for your use case.Define the role, recruit and set up the work.

Why bring in a PhD-led molecular design team?

A candidate ranking is useful when your scientists can see how it was made and challenge the assumptions before committing to synthesis. Our PhD-led team documents why designs were prioritised, where the predictions are uncertain and what experimental evidence would change the next round.

  • Agree on priorities. Tie the design and ranking criteria to the target and the next experiment.
  • Make the ranking reviewable. Describe the evidence and trade-offs behind each shortlisted design.
  • Use the assay results. Revisit the design assumptions when the lab returns new measurements.

Alex Gurbych

  • Founder and CEO · PhD in Computer Science

Alex co-authored peer-reviewed work on molecular binding affinity prediction and on predicting chemical reaction yields.

Binding affinity paper · Reaction yield paper · Our publications

What do clients say about working with Blackthorn AI?

Blackthorn AI has a company-wide rating of 5.0 from 22 reviews on Clutch. In life sciences, clients describe delivery of a working prototype, experience with scientific workflows and support with evaluating AI systems. Read the full reviews for the context and scope of each engagement.

Read all 22 reviews on Clutch

  • A biotech scientist describes receiving a working prototype for a research tool and valued the team’s initiative.

    Scientist, Biotech company

  • The client highlighted the team’s combined software and discovery expertise on a project involving a data warehouse, pipelines and a research portal.

    Diana Kraskouskaya, CEO, Dalriada Drug Discovery

  • The client valued the team’s technical expertise and communication during an AI consultation covering data analysis, training and validation.

    Yoanna Gouchtchina, CEO, ONE.SKIN Technologies

FAQ

What else should you know before starting a design round?

The feasibility check establishes which design options your inputs can support and what the next round would cost. These questions cover starting data, candidate files, screening criteria, lab responsibilities and how previous test results can inform the next set of molecules you plan to make.

01 What is AI molecular design?

AI molecular design uses computational models to propose and compare candidate structures or sequences. We use it to choose candidates for synthesis and experimental testing against criteria agreed with your scientists.

02 Can you start with a target and no existing leads?

Yes, if the available target information and intended assay support a credible design brief. The feasibility check assesses what can be designed and identifies missing evidence that could affect the choice.

03 Can you improve a lead we already have?

Yes. We can start from your structure or sequence and design variants around an agreed goal, such as predicted stability or selectivity. Assay data help us decide which trade-offs to explore.

04 What happens in the feasibility check?

We review the target, starting evidence, design criteria and experimental constraints. You get a written recommendation on whether to proceed, a proposed scope for the candidate design round and a quote.

05 How much does the feasibility check cost?

The fixed-fee feasibility check starts from €5,000 / £4,700 / $6,000 and takes 2–3 weeks. We agree the scope and fee before it starts.

06 How long does it take to receive a shortlist?

We quote the design-round schedule after feasibility. It depends on the modality, starting evidence, ranking criteria and handoff requirements. Synthesis and assay turnaround depend on the lab or CRO.

07 What do you score when ranking candidates?

Scoring depends on the brief. We can compare relevant predicted properties such as binding, selectivity or stability alongside practical constraints such as synthesis feasibility. We explain the criteria used and how uncertainty affects the order.

08 What files do we receive?

You receive candidate structures for small molecules or sequences for peptides, proteins and aptamers, plus the ranking rationale and lab handoff notes. We agree file formats and any predicted structural data in the scope.

09 How many candidates are on the shortlist?

We agree the batch size from the design scope and what your lab can test. The strength of the evidence may limit how many candidates we recommend. We include backups when they add useful options.

10 Who synthesises and tests the candidates?

Your lab or CRO runs synthesis and experiments. Blackthorn AI hands over the candidate specifications, recommends a testing order and identifies the predictions the experiments can help resolve.

11 What if no candidate meets the design criteria?

We document the weak or ambiguous results and what drove the decision. Depending on the evidence, we may propose a revised brief, request additional data or recommend pausing before you spend on synthesis.

12 Does a high score confirm the candidate will work?

A score is a computational estimate used to rank candidates. Your lab needs to measure the properties that matter, including activity or binding, and determine whether they meet the agreed criteria.

13 Can you use our experimental results in another round?

Yes. When you share the results, we can review where the predictions held up, identify useful changes and design and rank variants for the next synthesis batch.

14 How should we share proprietary data?

Bring a summary to the first call. We can discuss an NDA, access and an agreed transfer method before you send proprietary structures, sequences or assay data.

Which candidates do you need to design next?

Bring your target, an existing lead or a summary of the results from your last screen. We’ll discuss what you want the molecule to do, what your lab can test, and what a feasibility check could establish before you commit to a full design round.