MOTIVATION LETTER Antimicrobial resistance is a present, measurable crisis for Sub-Saharan Africa, not an abstract threat. The WHO African Region carries the world's highest burden of AMR-attributable deaths, while the continent remains the least served by genomic surveillance infrastructure. My research addresses this gap directly. I am Eniola Ayodele Olutogun, a Nigerian pharmacist and computational researcher, and I am applying to the L'Oréal-UNESCO For Women in Science Sub-Saharan Africa programme because my current work, the TOPOLOGIX project, builds a sequence-based prediction tool for drug-resistance mutations that works where structure-based tools fail, which is precisely the African context. TOPOLOGIX uses ESM-2 protein language model delta-embeddings combined with Morgan/ECFP drug fingerprints and a Random Forest classifier to predict drug-resistance mutations from sequence alone. On the Platinum benchmark of 553 mutations, it achieves an AUROC of 0.804 with a standard deviation of 0.025. On SKEMPI 2.0, it reaches 0.634. These numbers matter because they beat structure-based baselines such as mCSM-lig at approximately 0.70 while covering 100 percent of mutations, where structure-limited tools cover only about 18 percent. Most resistance-relevant proteins in African clinical isolates have no experimentally resolved structure. A sequence-only method removes that dependency entirely. This project is the direct result of a falsified hypothesis. My earlier pre-registered study on cardiotoxicity topology tested whether bipartite persistent homology could predict hERG cardiotoxicity from protein-ligand interface geometry. The powered replication found topological features do not beat a plain descriptor baseline, AUROC 0.8426 versus 0.8782. A second study applying the same topological constructs to drug-resistance prediction found almost no signal, AUROC 0.425 and 0.485 on the Platinum benchmark. Those negative results settled questions the literature had never actually run, and they redirected my work toward sequence representations. TOPOLOGIX is the product of that redirection. The L'Oréal-UNESCO programme's emphasis on technological sciences and on research that addresses real-world problems in Sub-Saharan Africa matches this project's design. The programme's evaluation criteria include scientific excellence, feasibility, and potential impact on society. TOPOLOGIX is feasible: it runs on standard HPC infrastructure, uses open datasets, and produces predictions that can be validated in any lab with sequencing capacity. Its societal impact is direct: knowing which resistance mutations are likely to emerge from a given protein sequence can inform drug selection and surveillance priorities across the region. I am a woman scientist working in computational pharmacology, a field where African women are severely underrepresented. I hold a B.Pharm from the University of Ibadan with a German-equivalent grade of 1.9, I am licensed by the Pharmacists Council of Nigeria, and I am enrolled in the M.Sc. Digital Health programme at the Hasso Plattner Institute and University of Potsdam. My collaborators include Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. I am applying at the pre-doctoral stage, which the programme's expanded eligibility now accommodates. The L'Oréal-UNESCO fellowship would fund the next validation phase of TOPOLOGIX: expanding the training set with African clinical resistance data and building a lightweight prediction service usable by public-health laboratories with limited compute. That is the concrete deliverable. The science is ready; the support is what I seek. RESEARCH STATEMENT Project title: TOPOLOGIX, sequence-based prediction of drug-resistance mutations for antimicrobial resistance surveillance in Sub-Saharan Africa. Problem and context. Antimicrobial resistance kills an estimated 1.27 million people globally per year, with Sub-Saharan Africa bearing a disproportionate share. Genomic surveillance is the cornerstone of AMR monitoring, but its utility depends on interpreting mutations in resistance-related proteins. Existing computational tools such as mCSM-lig require experimentally resolved protein structures. In African clinical settings, the majority of relevant proteins lack such structures. This creates a blind spot: the regions with the highest AMR burden are the least able to predict and interpret resistance mutations. TOPOLOGIX removes the structure requirement entirely by predicting resistance from amino-acid sequence alone. Methods. TOPOLOGIX combines three components. First, ESM-2 protein language model delta-embeddings capture the evolutionary and biophysical context of each mutation site from sequence. Second, Morgan/ECFP drug fingerprints encode the chemical identity of the ligand. Third, a Random Forest classifier integrates both representations to predict whether a given mutation confers resistance to a given drug. The pipeline is implemented in Python using scikit-learn and standard HPC tooling including Nextflow and SLURM, making it portable across institutional clusters. Results to date. On the Platinum benchmark, 553 mutations, TOPOLOGIX achieves AUROC 0.804 with a standard deviation of 0.025. On SKEMPI 2.0, it achieves 0.634. These results outperform structure-based baselines, mCSM-lig at approximately 0.70, while covering 100 percent of mutations in the benchmark. Structure-limited tools cover only about 18 percent of the same mutations. The performance gap is largest precisely for mutations in proteins without resolved structures, which is the target use case for African surveillance. Scientific rigor and provenance. TOPOLOGIX is the third project in a deliberate sequence. The first, a pre-registered, powered replication study, tested whether bipartite persistent homology predicts hERG cardiotoxicity from protein-ligand interface geometry. It found topological features do not beat a plain descriptor baseline, AUROC 0.8426 versus 0.8782. The second applied the same topological constructs to drug-resistance prediction and found no signal, AUROC 0.425 and 0.485 on the Platinum benchmark. These negative results, reported directly rather than reframed, ruled out interface geometry as the driver and motivated the sequence-representation approach that became TOPOLOGIX. The project's methods and validation gates are pre-registered. Proposed work for the fellowship period. Three objectives. First, expand the training set by incorporating resistance mutation data from African clinical isolates, particularly for Mycobacterium tuberculosis and Klebsiella pneumoniae, the two highest-burden AMR pathogens in the region. Second, validate TOPOLOGIX predictions against whole-genome sequencing data from published African AMR surveillance cohorts, measuring AUROC and calibration on held-out isolates. Third, package the trained model as a lightweight prediction service, a command-line tool and a simple web interface, deployable on a standard Linux VPS, so that public-health laboratories with minimal compute can query resistance predictions for novel mutations. Feasibility. All datasets are public. The Platinum benchmark and SKEMPI 2.0 are already integrated. African clinical isolate data is available through NCBI BioProject and the Pathogen Detection pipeline. Compute requirements are modest: ESM-2 embedding generation is a one-time cost per protein, and Random Forest training runs on a single node. I have operated HPC workflows on SLURM clusters and maintain production Linux systems with automated backup and disaster recovery. The infrastructure risk is low. Alignment with L'Oréal-UNESCO criteria. The programme seeks scientific excellence, feasibility, and societal impact in technological sciences. TOPOLOGIX demonstrates excellence through benchmarked, reproducible results and a documented history of falsification-driven method selection. It is feasible within a one-year fellowship timeline. Its societal impact is direct and measurable: improved resistance prediction for African pathogens, deployable in laboratories that currently have no predictive tool at all. The project also advances the programme's commitment to women in science by supporting a Nigerian woman researcher in computational pharmacology, a field where African women are underrepresented at every career stage. Career development plan. The fellowship period will produce three deliverables: a peer-reviewed publication on the expanded TOPOLOGIX model, an open-source software release with documentation, and a validated prediction service deployed in at least one collaborating African laboratory. This work will form the empirical core of my doctoral application, which I intend to submit to European and African doctoral programmes in computational biology and digital health during the fellowship year. ESSAY: LEADERSHIP AND COMMITMENT TO ADVANCING SCIENCE IN AFRICA My commitment to advancing science in Africa is demonstrated by what I have built as an independent researcher with no institutional laboratory backing. Between 2024 and 2026, I designed, executed, and reported five pre-registered computational research projects across addiction neuroscience, protein ML, and dynamical-systems methods. Each project required me to secure my own compute, manage my own literature reviews, and hold myself to publication standards without a supervisor's oversight. The largest literature screen covered 1,847 records to elicit Bayesian priors for the CCT model. That independence is evidence of the self-directed research capacity that African science needs. Leadership in this context means building infrastructure others can use. My four independent DuckDB-based ingest-to-analyze pipelines across life-sciences, tech, and social-science domains are designed to be reused. My neurocascade simulation engine has 62 passing tests and is structured so that other researchers can extend it to new receptor systems. TOPOLOGIX is being built as an open-source tool with a documented command-line interface, not a private research artifact. When I complete the M.Sc. Digital Health at the Hasso Plattner Institute and University of Potsdam, I intend to return to Nigeria and establish a computational biology group that trains local students in sequence-based ML methods, using TOPOLOGIX as the entry point. The L'Oréal-UNESCO programme's regional focus matters to me because it recognizes that African researchers face constraints that are not captured by publication counts alone. My work has been reviewed by collaborators at Michigan, Harvard, Princeton, and NYU, but I have done it without a dedicated lab, without grant funding, and without a local supervisor. A fellowship from this programme would change that trajectory. It would provide not only funding but also the formal recognition that enables institutional partnerships and doctoral admissions. I am asking for support to continue doing what I have already proven I can do, but with the resources to do it at scale and to bring others along. CHECKLIST - [ ] Confirm current eligibility for L'Oréal-UNESCO For Women in Science Sub-Saharan Africa: verify whether the programme requires doctoral or postdoctoral status, or whether pre-doctoral applicants are accepted in the current cycle - [ ] Verify that the programme accepts applicants enrolled in a European M.Sc. programme while conducting research relevant to Sub-Saharan Africa, or whether residence in a Sub-Saharan African country is required - [ ] Confirm the application deadline from the official L'Oréal-UNESCO For Women in Science website, not the engineerit.co.za article - [ ] Prepare CV in the programme's required format, including ORCID 0009-0001-9272-6735, GitHub, and zyco.org - [ ] Obtain two letters of recommendation: one from a named collaborator (Kent Berridge, Samuel Gershman, Nathaniel Daw, or Marcelo Mattar), one from a professional supervisor (Synthcare or Ramset Pharmacy) - [ ] Write a 250-word plain-language summary of TOPOLOGIX for non-specialist reviewers - [ ] Prepare a one-page budget breakdown for the fellowship period: compute costs, dataset access, publication fees, travel to a collaborating African laboratory - [ ] Prepare a timeline Gantt chart for the 12-month fellowship period with milestones for dataset expansion, validation, and software release - [ ] Verify the Platinum benchmark and SKEMPI 2.0 AUROC numbers against the current preprint versions before submission - [ ] Confirm the exact word limits for each application section on the official portal; adjust this draft to match if limits differ - [ ] Check whether the programme requires the research project to be conducted within a Sub-Saharan African institution; if so, identify a host laboratory in Nigeria and secure a letter of support - [ ] Prepare a data-management and open-science statement covering pre-registration, code release, and dataset deposition EDITOR NOTES - Eligibility risk is the primary concern: the deep-research summary states the programme historically requires doctoral or postdoctoral level, but the applicant is pre-doctoral (enrolled M.Sc., Winter 2026/27). The article title says "expand support," which may indicate a widened eligibility window, but this must be verified on the official portal before submission. If pre-doctoral applicants are excluded, this application should not proceed. - The applicant is a Nigerian national but is enrolled in a German M.Sc. programme and employed by Synthcare. The programme's residence requirement, if any, must be checked. If the programme requires research to be conducted in Sub-Saharan Africa, the applicant needs a host institution in Nigeria and a revised plan for remote or hybrid execution. - The TOPOLOGIX AUROC figures (0.804 on Platinum, 0.634 on SKEMPI 2.0) are taken from the applicant profile and should be re-verified against the current preprint versions before submission, as benchmarks may have been updated. - The motivation letter and research statement reference "expanded eligibility" and "pre-doctoral stage" as facts. These are inferences from the article title and should be softened or removed if the official programme description does not confirm them. - The applicant must insert personal details not present in this profile: specific reasons for choosing L'Oréal-UNESCO over other fellowships, any prior mentorship or outreach activities, and the name of the collaborating African laboratory for the deployment phase. These gaps are marked implicitly by their absence and must be filled by the applicant. - The research statement claims TOPOLOGIX can be validated "in any lab with sequencing capacity." This is accurate for the prediction service but does not address sample collection or sequencing costs. The budget section should clarify that the applicant is not proposing to generate new sequencing data, only to use existing public datasets. - The essay on leadership is the weakest section because the profile contains no explicit leadership or mentorship activities. The applicant should add concrete examples of teaching, mentoring, or community-building if any exist, or revise the essay to focus more narrowly on infrastructure-building as leadership.