Draft — Ocean State Labs

MOTIVATION LETTER

A protein language model that predicts drug resistance mutations from sequence alone, without requiring a crystal structure, can cover 100 percent of clinically observed mutations. Structure-dependent tools cover roughly 18 percent. On the Platinum benchmark, the venture achieves an AUROC of 0.804 plus or minus 0.025 using ESM-2 delta-embeddings combined with ECFP4 drug fingerprints and a Random Forest classifier. This beats the published SOTA, mCSM-lig, which reports an AUROC near 0.70. The venture is pre-seed, proof-of-concept validated, and not yet incorporated. I am Eniola Olutogun, a pharmacist turned ML engineer, and the sole founder.

Ocean State Labs is the first life sciences incubator in Rhode Island. Its network of hospitals and research institutes offers wet-lab validation partners that the venture needs to test resistance predictions in vitro. The incubator's mentorship from pharma executives can accelerate the SKEMPI 3K fine-tuning target, which aims to push AUROC above 0.70 on that benchmark, and the planned pilot with Servier in Suresnes. Rhode Island's growing biotech ecosystem, combined with access to specialized equipment and regulatory guidance, provides a practical bridge between computational prediction and clinical validation. The venture's global health impact spans antimicrobial resistance, oncology, and antiviral discovery. My background as a pharmacist who transitioned into machine learning engineering is a differentiator in a field dominated by PhD-only teams. I understand both the biological mechanism of resistance and the computational methods to predict it.

Ocean State Labs can provide the resources to move from a proof-of-concept to a revenue-generating platform. The venture targets an ARR model through pharma partnerships, starting with Servier. The incubator's support for early-stage life sciences startups aligns with the venture's need for lab space, industry connections, and mentorship. I am applying to Ocean State Labs to access these resources and to establish a US presence that complements the venture's primary EU focus in Ile-de-France.

SHORT ESSAY: SCIENTIFIC AND TECHNICAL MERIT

The venture's core innovation is a computational method that predicts drug resistance mutations from protein sequence alone. Most tools require a crystal structure, which limits coverage to approximately 18 percent of mutations. The venture's method uses ESM-2 protein language model delta-embeddings to capture sequence-level features, then combines them with ECFP4 drug fingerprints in a Random Forest classifier. On the Platinum benchmark of 553 mutations, using protein-grouped cross-validation, the model achieves an AUROC of 0.804 with a standard deviation of 0.025. On SKEMPI 2.0, the AUROC is 0.634. The published SOTA, mCSM-lig, reports an AUROC near 0.70 on comparable benchmarks. The venture's method already exceeds that performance on Platinum and covers all mutations, not just those with solved structures.

The technical roadmap focuses on fine-tuning ESM-2 on the SKEMPI 3K mutation dataset to push SKEMPI 2.0 AUROC above 0.70. This improvement will make the tool competitive for pharma validation. The venture has established relationships with Servier in Suresnes, Paris-Saclay's I2BC, Institut Pasteur, and Sanofi in Gentilly. These partners can provide wet-lab validation data and pilot opportunities. The scientific merit lies in solving a fundamental bottleneck in drug discovery: predicting resistance before it emerges clinically, which reduces late-stage drug failure and guides combination therapy design.

SHORT ESSAY: MARKET OPPORTUNITY AND COMPETITIVE LANDSCAPE

The global drug resistance testing market is projected to exceed 5 billion USD by 2030, driven by antimicrobial resistance, oncology resistance, and antiviral resistance. Current tools like mCSM-lig, Rosetta, and FoldX require crystal structures, limiting their use to well-characterized proteins. The venture's sequence-only approach addresses a gap: approximately 80 percent of clinically relevant mutations occur in proteins without solved structures. This includes many targets in oncology and infectious disease.

Competitors include companies like Recursion Pharmaceuticals, which uses high-throughput cell imaging and machine learning, and Insilico Medicine, which focuses on generative AI for drug design. Neither specializes in resistance prediction from sequence alone. The venture's unique value proposition is coverage: 100 percent of mutations versus 18 percent for structure-limited tools. The business model is a software-as-a-service platform sold to pharma companies, with an initial pilot at Servier. The venture targets an ARR of 500,000 euros within 24 months of incorporation, based on a per-target pricing model. Ocean State Labs can provide access to Rhode Island's hospital networks for clinical validation data, strengthening the commercial case.

SHORT ESSAY: TEAM QUALIFICATIONS AND DIVERSITY

I am Eniola Olutogun, a pharmacist licensed in Nigeria and a machine learning engineer. I hold a degree in pharmacy and have completed advanced coursework in deep learning and computational biology. My professional experience includes clinical pharmacy practice, where I observed drug resistance failures firsthand, and subsequent transition into ML engineering, where I built the venture's prediction pipeline as a sole author. This dual background is rare: most teams in computational drug discovery are led by PhDs in computational biology or chemistry who lack clinical experience. I understand the patient-level consequences of resistance and the data structures needed to predict it.

As a Nigerian woman founder in biotech AI, I bring a perspective underrepresented in the European and US startup ecosystems. Ocean State Labs values diversity in its cohort. My trajectory from pharmacist to ML engineer to founder demonstrates adaptability and technical depth. I have submitted applications to SEMIA and Quest for Health, WILCO One BioTech in October 2026, and IncubAlliance plus AI House. The EIC Accelerator and BPI i-Lab are future targets. Ocean State Labs can provide the US foothold and mentorship to complement these EU-focused programs.

SHORT ESSAY: FEASIBILITY OF MILESTONES AND USE OF INCUBATOR RESOURCES

The venture's 12-month milestones are: fine-tune ESM-2 on SKEMPI 3K mutations to achieve AUROC above 0.70 on SKEMPI 2.0, complete the Servier pilot agreement, and generate first revenue from a pharma partnership. Ocean State Labs resources will be used for wet-lab validation space to test predictions in vitro, mentorship from pharma executives to refine the commercial strategy, and access to Rhode Island hospital networks for clinical resistance data. The incubator's specialized equipment, if available, can support in vitro resistance assays.

The 24-month milestone is an ARR of 500,000 euros from 3 to 5 pharma clients. The venture will use Ocean State Labs' network to identify and secure these clients. The incubator's location in the US provides a strategic complement to the EU focus, allowing the venture to serve both markets. The founder's pharmacist background ensures realistic assessment of validation timelines. The venture is not yet incorporated, but incorporation in France or the US will occur within three months of acceptance into an incubator program.

RESEARCH STATEMENT

The venture addresses a specific problem in drug discovery: predicting which mutations in a protein will confer resistance to a given drug, using only the protein's amino acid sequence. Current methods require a three-dimensional crystal structure, which is unavailable for approximately 82 percent of clinically relevant mutations. This structural bottleneck means that resistance predictions are limited to well-studied proteins, leaving most resistance mechanisms uncharacterized until they emerge in patients.

The technical approach uses ESM-2, a protein language model trained on 250 million protein sequences, to generate delta-embeddings that capture the effect of a mutation on the protein's representation. These embeddings are combined with ECFP4 drug fingerprints, which encode the chemical structure of the drug, and fed into a Random Forest classifier. The model outputs a probability of resistance for each mutation-drug pair. On the Platinum benchmark, which contains 553 mutations across 10 proteins, the model achieves an AUROC of 0.804 with protein-grouped cross-validation. On SKEMPI 2.0, a more challenging benchmark of binding affinity changes, the AUROC is 0.634. The published SOTA, mCSM-lig, reports an AUROC near 0.70 on comparable benchmarks.

The immediate research goal is to fine-tune ESM-2 on the SKEMPI 3K dataset, which contains approximately 3,000 mutations with binding affinity data. This fine-tuning is expected to push SKEMPI 2.0 AUROC above 0.70, making the tool competitive for pharma validation. The venture has established relationships with Servier, Paris-Saclay, Institut Pasteur, and Sanofi for wet-lab validation and pilot studies. The long-term research goal is to extend the method to predict resistance in combination therapies and to incorporate clinical outcome data from hospital networks, such as those accessible through Ocean State Labs in Rhode Island.

The broader impact is on antimicrobial resistance, oncology resistance, and antiviral resistance. Predicting resistance before it emerges allows drug developers to design compounds that evade common resistance mechanisms, select combination therapies proactively, and reduce late-stage clinical trial failures. The venture's platform is computational and dry-lab, but requires wet-lab validation partners to confirm predictions. Ocean State Labs can provide access to such partners through its Rhode Island hospital and research institute network.

CHECKLIST

- [ ] Motivation letter, 300-500 words, tailored to Ocean State Labs
- [ ] Short essay on scientific and technical merit, 200-350 words
- [ ] Short essay on market opportunity and competitive landscape, 200-350 words
- [ ] Short essay on team qualifications and diversity, 200-350 words
- [ ] Short essay on feasibility of milestones and use of incubator resources, 200-350 words
- [ ] Research statement, 400-600 words
- [ ] Founder CV or resume
- [ ] Proof-of-concept validation data summary (AUROC 0.804 on Platinum, 0.634 on SKEMPI 2.0)
- [ ] Letters of support or intent from named partners (Servier, Paris-Saclay, Institut Pasteur, Sanofi) if available
- [ ] Incorporation plan or timeline
- [ ] Financial projection or budget for 12-24 months
- [ ] Any additional documents required by Ocean State Labs application portal

EDITOR NOTES

- Eligibility risk: Ocean State Labs is a US incubator. The venture's primary geography is EU, specifically France and Ile-de-France. Confirm that the incubator accepts international founders or requires US incorporation. If required, plan for US incorporation timeline.
- Fact verification: Confirm the Platinum benchmark AUROC of 0.804 plus or minus 0.025 and SKEMPI 2.0 AUROC of 0.634 are from published or preprint results. If not yet published, note that in the application.
- Gap: The profile does not specify the founder's current location or visa status. Ocean State Labs may require the founder to relocate to Rhode Island for the incubator duration. Insert this detail if known.
- Gap: The profile mentions SEMIA, Quest for Health, WILCO One BioTech, IncubAlliance, AI House, EIC Accelerator, and BPI i-Lab as support pipeline. Confirm which applications are submitted and which are pending. Ocean State Labs may ask about other incubator commitments.
- Gap: The venture is not yet incorporated. Ocean State Labs may require a legal entity to accept funding or resources. Include a plan for incorporation within three months of acceptance.

MOTIVATION LETTER

A protein language model that predicts drug resistance mutations from sequence alone, without requiring a crystal structure, can cover 100 percent of clinically observed mutations. Structure-dependent tools cover roughly 18 percent. On the Platinum benchmark, the venture achieves an AUROC of 0.804 plus or minus 0.025 using ESM-2 delta-embeddings combined with ECFP4 drug fingerprints and a Random Forest classifier. This beats the published SOTA, mCSM-lig, which reports an AUROC near 0.70. The venture is pre-seed, proof-of-concept validated, and not yet incorporated. I am Eniola Olutogun, a pharmacist turned ML engineer, and the sole founder.

Ocean State Labs is the first life sciences incubator in Rhode Island. Its network of hospitals and research institutes offers wet-lab validation partners that the venture needs to test resistance predictions in vitro. The incubator's mentorship from pharma executives can accelerate the SKEMPI 3K fine-tuning target, which aims to push AUROC above 0.70 on that benchmark, and the planned pilot with Servier in Suresnes. Rhode Island's growing biotech ecosystem, combined with access to specialized equipment and regulatory guidance, provides a practical bridge between computational prediction and clinical validation. The venture's global health impact spans antimicrobial resistance, oncology, and antiviral discovery. My background as a pharmacist who transitioned into machine learning engineering is a differentiator in a field dominated by PhD-only teams. I understand both the biological mechanism of resistance and the computational methods to predict it.

Ocean State Labs can provide the resources to move from a proof-of-concept to a revenue-generating platform. The venture targets an ARR model through pharma partnerships, starting with Servier. The incubator's support for early-stage life sciences startups aligns with the venture's need for lab space, industry connections, and mentorship. I am applying to Ocean State Labs to access these resources and to establish a US presence that complements the venture's primary EU focus in Ile-de-France.

SHORT ESSAY: SCIENTIFIC AND TECHNICAL MERIT

The venture's core innovation is a computational method that predicts drug resistance mutations from protein sequence alone. Most tools require a crystal structure, which limits coverage to approximately 18 percent of mutations. The venture's method uses ESM-2 protein language model delta-embeddings to capture sequence-level features, then combines them with ECFP4 drug fingerprints in a Random Forest classifier. On the Platinum benchmark of 553 mutations, using protein-grouped cross-validation, the model achieves an AUROC of 0.804 with a standard deviation of 0.025. On SKEMPI 2.0, the AUROC is 0.634. The published SOTA, mCSM-lig, reports an AUROC near 0.70 on comparable benchmarks. The venture's method already exceeds that performance on Platinum and covers all mutations, not just those with solved structures.

The technical roadmap focuses on fine-tuning ESM-2 on the SKEMPI 3K mutation dataset to push SKEMPI 2.0 AUROC above 0.70. This improvement will make the tool competitive for pharma validation. The venture has established relationships with Servier in Suresnes, Paris-Saclay's I2BC, Institut Pasteur, and Sanofi in Gentilly. These partners can provide wet-lab validation data and pilot opportunities. The scientific merit lies in solving a fundamental bottleneck in drug discovery: predicting resistance before it emerges clinically, which reduces late-stage drug failure and guides combination therapy design.

SHORT ESSAY: MARKET OPPORTUNITY AND COMPETITIVE LANDSCAPE

The global drug resistance testing market is projected to exceed 5 billion USD by 2030, driven by antimicrobial resistance, oncology resistance, and antiviral resistance. Current tools like mCSM-lig, Rosetta, and FoldX require crystal structures, limiting their use to well-characterized proteins. The venture's sequence-only approach addresses a gap: approximately 80 percent of clinically relevant mutations occur in proteins without solved structures. This includes many targets in oncology and infectious disease.

Competitors include companies like Recursion Pharmaceuticals, which uses high-throughput cell imaging and machine learning, and Insilico Medicine, which focuses on generative AI for drug design. Neither specializes in resistance prediction from sequence alone. The venture's unique value proposition is coverage: 100 percent of mutations versus 18 percent for structure-limited tools. The business model is a software-as-a-service platform sold to pharma companies, with an initial pilot at Servier. The venture targets an ARR of 500,000 euros within 24 months of incorporation, based on a per-target pricing model. Ocean State Labs can provide access to Rhode Island's hospital networks for clinical validation data, strengthening the commercial case.

SHORT ESSAY: TEAM QUALIFICATIONS AND DIVERSITY

I am Eniola Olutogun, a pharmacist licensed in Nigeria and a machine learning engineer. I hold a degree in pharmacy and have completed advanced coursework in deep learning and computational biology. My professional experience includes clinical pharmacy practice, where I observed drug resistance failures firsthand, and subsequent transition into ML engineering, where I built the venture's prediction pipeline as a sole author. This dual background is rare: most teams in computational drug discovery are led by PhDs in computational biology or chemistry who lack clinical experience. I understand the patient-level consequences of resistance and the data structures needed to predict it.

As a Nigerian woman founder in biotech AI, I bring a perspective underrepresented in the European and US startup ecosystems. Ocean State Labs values diversity in its cohort. My trajectory from pharmacist to ML engineer to founder demonstrates adaptability and technical depth. I have submitted applications to SEMIA and Quest for Health, WILCO One BioTech in October 2026, and IncubAlliance plus AI House. The EIC Accelerator and BPI i-Lab are future targets. Ocean State Labs can provide the US foothold and mentorship to complement these EU-focused programs.

SHORT ESSAY: FEASIBILITY OF MILESTONES AND USE OF INCUBATOR RESOURCES

The venture's 12-month milestones are: fine-tune ESM-2 on SKEMPI 3K mutations to achieve AUROC above 0.70 on SKEMPI 2.0, complete the Servier pilot agreement, and generate first revenue from a pharma partnership. Ocean State Labs resources will be used for wet-lab validation space to test predictions in vitro, mentorship from pharma executives to refine the commercial strategy, and access to Rhode Island hospital networks for clinical resistance data. The incubator's specialized equipment, if available, can support in vitro resistance assays.

The 24-month milestone is an ARR of 500,000 euros from 3 to 5 pharma clients. The venture will use Ocean State Labs' network to identify and secure these clients. The incubator's location in the US provides a strategic complement to the EU focus, allowing the venture to serve both markets. The founder's pharmacist background ensures realistic assessment of validation timelines. The venture is not yet incorporated, but incorporation in France or the US will occur within three months of acceptance into an incubator program.

RESEARCH STATEMENT

The venture addresses a specific problem in drug discovery: predicting which mutations in a protein will confer resistance to a given drug, using only the protein's amino acid sequence. Current methods require a three-dimensional crystal structure, which is unavailable for approximately 82 percent of clinically relevant mutations. This structural bottleneck means that resistance predictions are limited to well-studied proteins, leaving most resistance mechanisms uncharacterized until they emerge in patients.

The technical approach uses ESM-2, a protein language model trained on 250 million protein sequences, to generate delta-embeddings that capture the effect of a mutation on the protein's representation. These embeddings are combined with ECFP4 drug fingerprints, which encode the chemical structure of the drug, and fed into a Random Forest classifier. The model outputs a probability of resistance for each mutation-drug pair. On the Platinum benchmark, which contains 553 mutations across 10 proteins, the model achieves an AUROC of 0.804 with protein-grouped cross-validation. On SKEMPI 2.0, a more challenging benchmark of binding affinity changes, the AUROC is 0.634. The published SOTA, mCSM-lig, reports an AUROC near 0.70 on comparable benchmarks.

The immediate research goal is to fine-tune ESM-2 on the SKEMPI 3K dataset, which contains approximately 3,000 mutations with binding affinity data. This fine-tuning is expected to push SKEMPI 2.0 AUROC above 0.70, making the tool competitive for pharma validation. The venture has established relationships with Servier, Paris-Saclay, Institut Pasteur, and Sanofi for wet-lab validation and pilot studies. The long-term research goal is to extend the method to predict resistance in combination therapies and to incorporate clinical outcome data from hospital networks, such as those accessible through Ocean State Labs in Rhode Island.

The broader impact is on antimicrobial resistance, oncology resistance, and antiviral resistance. Predicting resistance before it emerges allows drug developers to design compounds that evade common resistance mechanisms, select combination therapies proactively, and reduce late-stage clinical trial failures. The venture's platform is computational and dry-lab, but requires wet-lab validation partners to confirm predictions. Ocean State Labs can provide access to such partners through its Rhode Island hospital and research institute network.

CHECKLIST

- [ ] Motivation letter, 300-500 words, tailored to Ocean State Labs
- [ ] Short essay on scientific and technical merit, 200-350 words
- [ ] Short essay on market opportunity and competitive landscape, 200-350 words
- [ ] Short essay on team qualifications and diversity, 200-350 words
- [ ] Short essay on feasibility of milestones and use of incubator resources, 200-350 words
- [ ] Research statement, 400-600 words
- [ ] Founder CV or resume
- [ ] Proof-of-concept validation data summary (AUROC 0.804 on Platinum, 0.634 on SKEMPI 2.0)
- [ ] Letters of support or intent from named partners (Servier, Paris-Saclay, Institut Pasteur, Sanofi) if available
- [ ] Incorporation plan or timeline
- [ ] Financial projection or budget for 12-24 months
- [ ] Any additional documents required by Ocean State Labs application portal

EDITOR NOTES

- Eligibility risk: Ocean State Labs is a US incubator. The venture's primary geography is EU, specifically France and Ile-de-France. Confirm that the incubator accepts international founders or requires US incorporation. If required, plan for US incorporation timeline.
- Fact verification: Confirm the Platinum benchmark AUROC of 0.804 plus or minus 0.025 and SKEMPI 2.0 AUROC of 0.634 are from published or preprint results. If not yet published, note that in the application.
- Gap: The profile does not specify the founder's current location or visa status. Ocean State Labs may require the founder to relocate to Rhode Island for the incubator duration. Insert this detail if known.
- Gap: The profile mentions SEMIA, Quest for Health, WILCO One BioTech, IncubAlliance, AI House, EIC Accelerator, and BPI i-Lab as support pipeline. Confirm which applications are submitted and which are pending. Ocean State Labs may ask about other incubator commitments.
- Gap: The venture is not yet incorporated. Ocean State Labs may require a legal entity to accept funding or resources. Include a plan for incorporation within three months of acceptance.

AI Draft — Ocean State Labs