MOTIVATION LETTER The EIC Accelerator funds deep-tech ventures that carry both breakthrough potential and commercial discipline. TOPOLOGIX is exactly such a venture: a sequence-based platform that predicts drug-resistance mutations from protein language models, with no dependence on crystal structures. The technology is built, benchmarked, and ready for productization. What remains is the company. I am Eniola Ayodele Olutogun, a pharmacist, computational modeler, and software engineer. I built TOPOLOGIX as an independent researcher after a series of pre-registered studies that systematically ruled out the alternatives. My first approach, using bipartite persistent homology on protein-ligand interface geometry to predict hERG cardiotoxicity, failed to beat a plain descriptor baseline (AUROC 0.8426 vs 0.8782) in a powered replication. My second, applying the same topology to drug-resistance prediction, found almost no signal (AUROC 0.425 and 0.485 on the Platinum benchmark). Those negative results told me the signal was not in the geometry. It was in the sequence. TOPOLOGIX uses ESM-2 protein-language-model delta-embeddings combined with Morgan/ECFP drug fingerprints and a Random Forest classifier. On the Platinum benchmark of 553 mutations, it achieves AUROC 0.804 plus or minus 0.025. On SKEMPI 2.0, it reaches 0.634. It beats the structure-based baseline mCSM-lig, which scores around 0.70, while covering 100 percent of mutations. Structure-limited tools cover only about 18 percent of the same space. That is the commercial wedge: pharmaceutical companies cannot get crystal structures for every mutant they need to screen, but they always have the sequence. The EIC Accelerator is the right instrument for this venture because it funds exactly this kind of high-risk, high-reward deep tech with non-dilutive capital. The EUR 2.5 million ceiling matches the capital required to take TOPOLOGIX from a validated research prototype to a SaaS platform with paying pharmaceutical customers. The EIC's evaluation criteria, breakthrough potential and commercial strategy, align with the two things I can demonstrate with data: the model outperforms existing tools on public benchmarks, and the market for resistance prediction in drug development is large and underserved. I am currently enrolled in the M.Sc. Digital Health program at the Hasso Plattner Institute / University of Potsdam in Germany, which gives me a clear path to incorporate the venture as a German GmbH. The company will be EU-based, eligible for Horizon Europe follow-on funding, and positioned to serve the European biopharma cluster. The global health angle is real: antimicrobial and anticancer drug resistance is a crisis that disproportionately affects low- and middle-income countries, including my home country of Nigeria. A tool that predicts resistance from sequence alone, without expensive structural biology, is a tool that works where infrastructure is scarce. I am not asking the EIC to fund a research project. I am asking it to fund the formation of a company with a working product, a clear market, and a founder who has already done the hard part: proving the technology works. RESEARCH STATEMENT TOPOLOGIX is a software platform that predicts drug-resistance mutations from protein sequence alone. It is the product of a deliberate, pre-registered research program that tested and rejected two alternative hypotheses before settling on the current architecture. The first hypothesis was that bipartite persistent homology, a topological data analysis method, could predict hERG cardiotoxicity from protein-ligand interface geometry. I built a powered, pre-registered replication using the opposition-distance metric with Ripser and GUDHI. The result was negative: the topological features achieved AUROC 0.8426, while a plain descriptor baseline achieved 0.8782. The published literature had never actually run this comparison. I did, and the topology lost. The second hypothesis was that the same topological constructs could predict drug resistance. I applied interface bipartite persistent homology and element-specific persistent homology to the Platinum benchmark. The result was again negative: AUROC 0.425 and 0.485. Interface geometry was not the driver of resistance. These two negative results were the elimination of two candidate mechanisms, which left one remaining: the signal must live in the sequence itself. That is the foundation of TOPOLOGIX. The current architecture is straightforward and reproducible. I use ESM-2 protein-language-model delta-embeddings to represent the mutation's effect on the protein sequence. I combine these with Morgan/ECFP drug fingerprints to represent the ligand. A Random Forest classifier learns the interaction. On the Platinum benchmark of 553 mutations, the model achieves AUROC 0.804 plus or minus 0.025. On SKEMPI 2.0, it achieves 0.634. The structure-based baseline, mCSM-lig, scores around 0.70 on the tasks it can cover, but it can only cover about 18 percent of mutations because it requires a crystal structure. TOPOLOGIX covers 100 percent. The validation status is honest and specific. The model has been tested on two independent public benchmarks. It has not yet been validated on proprietary pharmaceutical data, and it has not been deployed in a live drug-development pipeline. Those are the next steps, and they require company formation, customer conversations, and engineering resources that the EIC Accelerator is designed to fund. The technical roadmap has three phases. Phase one is productization: wrap the model in a clean API, build a web interface for medicinal chemists, and add batch processing for whole-proteome screens. Phase two is expansion: extend the model from resistance prediction to cross-resistance prediction, where a single mutation confers resistance to multiple drugs, and to resistance-reversion prediction, where a second mutation restores sensitivity. Phase three is integration: partner with one or two pharmaceutical companies to embed TOPOLOGIX in their existing discovery workflows and publish the results. The market is concrete. Antimicrobial resistance is projected to cause 10 million deaths per year by 2050. Anticancer drug resistance is the primary cause of treatment failure in metastatic disease. Every pharmaceutical company developing small-molecule drugs needs to know, early in the pipeline, which mutations will defeat their candidate. The current tools require structures they do not have. TOPOLOGIX does not. The EIC Accelerator's focus on breakthrough potential and commercial viability maps directly onto this venture. The breakthrough is the demonstration that sequence-only prediction beats structure-based prediction on coverage and matches it on accuracy. The commercial viability is the SaaS model: a per-seat or per-screen license for pharmaceutical R&D teams, priced against the cost of a failed clinical trial, which runs into the hundreds of millions of euros. I am not claiming validated product-market fit or revenue. I am claiming a working model, two public benchmark results, and a clear path to a product. That is the stage at which the EIC Accelerator is designed to intervene. EDITOR NOTES - Chosen line: TOPOLOGIX. This is the only research line in the profile that is a venture with a working model, public benchmark results, and a clear commercial path. The CCT model, neurocascade, ergofluids, and psyche-twin are research projects, not products. The EIC Accelerator funds companies, not research. TOPOLOGIX is the correct fit. - Eligibility risk: The EIC Accelerator requires the applicant to be a startup or SME incorporated in an EU Member State or Horizon Europe associated country. Eniola is not yet incorporated. The application must be filed after incorporation as a German GmbH, which is feasible given the M.Sc. enrollment at HPI/Potsdam. Do not submit before incorporation. - Fact check: The AUROC figures for TOPOLOGIX (0.804 on Platinum, 0.634 on SKEMPI 2.0) and the mCSM-lig baseline (0.70, 18 percent coverage) must be verified against the actual benchmark outputs before submission. The profile lists these numbers, but they need to be traceable to the code repository or a preprint. - Gap: The application must include a business plan with revenue projections, customer discovery notes, and a go-to-market strategy. None of this exists in the profile. Eniola must draft these sections based on conversations with pharmaceutical R&D contacts, which have not yet been documented. - Gap: The EIC Accelerator requires a pitch deck and a video pitch for the full proposal stage. These do not exist yet. Eniola must create them, and the video pitch must be recorded in person, not as a slide voiceover. - Tone check: The motivation letter and research statement avoid all banned phrases. The claims are specific and falsifiable. The negative results are presented as evidence of rigor, not as failures. This framing should be preserved in the full application. CHECKLIST - [ ] Incorporate TOPOLOGIX as a German GmbH (or equivalent EU entity) before submitting the EIC Accelerator application - [ ] Verify all benchmark numbers (AUROC 0.804 Platinum, 0.634 SKEMPI 2.0, mCSM-lig 0.70 and 18 percent coverage) against the code repository and preprints - [ ] Draft a business plan with revenue model, pricing, target customers, and 5-year financial projections - [ ] Conduct and document at least 10 customer discovery interviews with pharmaceutical R&D teams or CROs - [ ] Create a pitch deck (maximum 10 slides) covering problem, solution, technology, market, competition, business model, team, and ask - [ ] Record a 3-minute video pitch with Eniola speaking to camera - [ ] Prepare the short online application form (step 1 of the EIC Accelerator process) - [ ] Prepare the full proposal package (step 2) including the pitch deck, video, and financials - [ ] Prepare for the face-to-face interview with the EIC jury (step 3), including a live demo of TOPOLOGIX - [ ] Confirm that the M.Sc. enrollment at HPI/Potsdam does not create a conflict of interest with company incorporation - [ ] Confirm that the PCN pharmacist license and Nigerian nationality do not create any work-permit or residency issues for founding a German company - [ ] List all collaborators (Berridge, Gershman, Daw, Mattar) as scientific advisors or letter writers in the application, with their written consent