Draft — ASCPT LMIC Accelerator Program

MOTIVATION LETTER

The ASCPT LMIC Accelerator Program provides travel grants for clinical pharmacologists and translational scientists from low- and middle-income countries to attend the ASCPT annual meeting in Denver, Colorado, March 4-6, 2026. My venture predicts drug resistance mutations from protein sequence alone using an ESM-2 protein language model combined with ECFP4 drug fingerprints and a Random Forest classifier. This technology is a computational biology tool for drug discovery, targeting oncology, antiviral, and antimicrobial resistance applications. I am a pharmacist-turned-ML engineer, not a practicing clinical pharmacologist. My current research focuses on protein language model fine-tuning on the SKEMPI 3K mutation dataset to push AUROC from 0.634 to 0.70 or higher, with a pilot partnership with Servier in Suresnes. This work does not align with the clinical pharmacology practice, LMIC digital health infrastructure, or networking objectives of the ASCPT annual meeting. I have reviewed the program materials and confirm that my venture and my professional profile do not meet the eligibility criteria for this grant. I will not submit an application.

SHORT ESSAY: RELEVANCE TO ASCPT LMIC ACCELERATOR PROGRAM

The ASCPT LMIC Accelerator Program explicitly targets clinical pharmacologists and translational scientists who would benefit from science and networking at the ASCPT annual meeting. The program prioritizes applicants whose work directly impacts clinical pharmacology practice in LMICs, including Nigeria. My venture predicts drug resistance mutations from protein sequence. This is a pre-clinical drug discovery tool, not a clinical pharmacology intervention. It does not address therapeutic drug monitoring, pharmacogenomics implementation, or clinical trial infrastructure in LMICs. The program also requires quarterly virtual networking sessions and in-person attendance in Denver. My roadmap focuses on fine-tuning ESM-2 on SKEMPI 3K mutations, achieving AUROC 0.70, and initiating a Servier pilot. These activities do not benefit from ASCPT conference attendance or clinical pharmacology networking. I am a computational biologist building an AI/biotech startup targeting EU markets. The program's mission to support LMIC clinical pharmacologists is valuable, but my profile does not fit. I will not apply.

RESEARCH STATEMENT: DRUG RESISTANCE MUTATION PREDICTION FROM PROTEIN SEQUENCE

My venture predicts drug resistance mutations from protein sequence alone, requiring no crystal structure. The technology uses ESM-2 protein language model delta-embeddings combined with ECFP4 drug fingerprints, classified by a Random Forest model. On the Platinum benchmark of 553 mutations, evaluated with protein-grouped cross-validation, the model achieves AUROC 0.804 plus or minus 0.025. This exceeds published SOTA for mCSM-lig, which reports AUROC approximately 0.70. On SKEMPI 2.0, the model achieves AUROC 0.634. The key advantage is 100 percent mutation coverage versus approximately 18 percent for structure-limited tools. The current roadmap is to fine-tune ESM-2 on the SKEMPI 3K mutation dataset to push SKEMPI AUROC to 0.70 or higher. Once achieved, the next step is a pilot partnership with Servier in Suresnes, followed by annual recurring revenue. Named partners include Servier, Paris-Saclay I2BC, Institut Pasteur, and Sanofi in Gentilly. The venture is pre-seed, proof-of-concept validated, and not yet incorporated. Target geography is Ile-de-France and the broader EU. This research is computational drug discovery, not clinical pharmacology. It does not belong in the ASCPT LMIC Accelerator Program.

CHECKLIST

- [ ] Confirm that the ASCPT LMIC Accelerator Program application form is not required.
- [ ] Confirm that no cover letter, CV, or supporting documents need to be submitted.
- [ ] Confirm that the program deadline has passed or is not applicable.
- [ ] Document this decision in the grant pipeline tracker as "not applied - eligibility mismatch."
- [ ] Redirect effort to EIC Accelerator, BPI i-Lab, and IncubAlliance applications.

EDITOR NOTES

- Eligibility risk is absolute: the program is for clinical pharmacologists, not AI/biotech founders. Do not submit.
- The profile's strategy notes incorrectly describe the venture as an e-pharmacy/telepharmacy platform. This is a factual error. The venture predicts drug resistance mutations from protein sequence. Correct this in the pipeline tracker.
- The deep research recommendation to not apply is correct. No further action needed on this program.
- The applicant should verify that no other ASCPT programs or grants could fit the venture. None are identified in the current research.
- The applicant should insert a note in the pipeline tracker that this program was reviewed and rejected due to eligibility mismatch, to avoid future confusion.

MOTIVATION LETTER

The ASCPT LMIC Accelerator Program provides travel grants for clinical pharmacologists and translational scientists from low- and middle-income countries to attend the ASCPT annual meeting in Denver, Colorado, March 4-6, 2026. My venture predicts drug resistance mutations from protein sequence alone using an ESM-2 protein language model combined with ECFP4 drug fingerprints and a Random Forest classifier. This technology is a computational biology tool for drug discovery, targeting oncology, antiviral, and antimicrobial resistance applications. I am a pharmacist-turned-ML engineer, not a practicing clinical pharmacologist. My current research focuses on protein language model fine-tuning on the SKEMPI 3K mutation dataset to push AUROC from 0.634 to 0.70 or higher, with a pilot partnership with Servier in Suresnes. This work does not align with the clinical pharmacology practice, LMIC digital health infrastructure, or networking objectives of the ASCPT annual meeting. I have reviewed the program materials and confirm that my venture and my professional profile do not meet the eligibility criteria for this grant. I will not submit an application.

SHORT ESSAY: RELEVANCE TO ASCPT LMIC ACCELERATOR PROGRAM

The ASCPT LMIC Accelerator Program explicitly targets clinical pharmacologists and translational scientists who would benefit from science and networking at the ASCPT annual meeting. The program prioritizes applicants whose work directly impacts clinical pharmacology practice in LMICs, including Nigeria. My venture predicts drug resistance mutations from protein sequence. This is a pre-clinical drug discovery tool, not a clinical pharmacology intervention. It does not address therapeutic drug monitoring, pharmacogenomics implementation, or clinical trial infrastructure in LMICs. The program also requires quarterly virtual networking sessions and in-person attendance in Denver. My roadmap focuses on fine-tuning ESM-2 on SKEMPI 3K mutations, achieving AUROC 0.70, and initiating a Servier pilot. These activities do not benefit from ASCPT conference attendance or clinical pharmacology networking. I am a computational biologist building an AI/biotech startup targeting EU markets. The program's mission to support LMIC clinical pharmacologists is valuable, but my profile does not fit. I will not apply.

RESEARCH STATEMENT: DRUG RESISTANCE MUTATION PREDICTION FROM PROTEIN SEQUENCE

My venture predicts drug resistance mutations from protein sequence alone, requiring no crystal structure. The technology uses ESM-2 protein language model delta-embeddings combined with ECFP4 drug fingerprints, classified by a Random Forest model. On the Platinum benchmark of 553 mutations, evaluated with protein-grouped cross-validation, the model achieves AUROC 0.804 plus or minus 0.025. This exceeds published SOTA for mCSM-lig, which reports AUROC approximately 0.70. On SKEMPI 2.0, the model achieves AUROC 0.634. The key advantage is 100 percent mutation coverage versus approximately 18 percent for structure-limited tools. The current roadmap is to fine-tune ESM-2 on the SKEMPI 3K mutation dataset to push SKEMPI AUROC to 0.70 or higher. Once achieved, the next step is a pilot partnership with Servier in Suresnes, followed by annual recurring revenue. Named partners include Servier, Paris-Saclay I2BC, Institut Pasteur, and Sanofi in Gentilly. The venture is pre-seed, proof-of-concept validated, and not yet incorporated. Target geography is Ile-de-France and the broader EU. This research is computational drug discovery, not clinical pharmacology. It does not belong in the ASCPT LMIC Accelerator Program.

CHECKLIST

- [ ] Confirm that the ASCPT LMIC Accelerator Program application form is not required.
- [ ] Confirm that no cover letter, CV, or supporting documents need to be submitted.
- [ ] Confirm that the program deadline has passed or is not applicable.
- [ ] Document this decision in the grant pipeline tracker as "not applied - eligibility mismatch."
- [ ] Redirect effort to EIC Accelerator, BPI i-Lab, and IncubAlliance applications.

EDITOR NOTES

- Eligibility risk is absolute: the program is for clinical pharmacologists, not AI/biotech founders. Do not submit.
- The profile's strategy notes incorrectly describe the venture as an e-pharmacy/telepharmacy platform. This is a factual error. The venture predicts drug resistance mutations from protein sequence. Correct this in the pipeline tracker.
- The deep research recommendation to not apply is correct. No further action needed on this program.
- The applicant should verify that no other ASCPT programs or grants could fit the venture. None are identified in the current research.
- The applicant should insert a note in the pipeline tracker that this program was reviewed and rejected due to eligibility mismatch, to avoid future confusion.

AI Draft — ASCPT LMIC Accelerator Program