MOTIVATION LETTER The CCT model I built and calibrated with Bayesian MCMC estimates causal effects of dopaminergic reward prediction error on reward-memory encoding in addiction. That model confirmed all five pre-registered hypotheses with posterior super-additivity of 13 to 22 percentage points across versions. But I built those causal claims using dynamical systems and pharmacological priors alone, without formal training in causal inference frameworks. The EMBO Practical Course on causality in biomedicine at EMBL-EBI directly addresses this gap. I am an independent computational researcher based in Nigeria, licensed as a pharmacist, enrolled in the M.Sc. Digital Health at Hasso Plattner Institute starting winter 2026/27. My research spans addiction neuroscience, protein-drug machine learning, and dynamical-systems methods. I have sole-authored three preprints on the CCT model, co-authored a paper under review at Alcohol (Elsevier), and built a drug-resistance mutation predictor (TOPOLOGIX) that achieves AUROC 0.804 on the Platinum benchmark, beating structure-based tools that cover only 18 percent of mutations. The course will equip me to formalize the causal reasoning already embedded in my work. My CCT model uses a three-axis ODE system with Bayesian calibration to infer how dopaminergic RPE, NMDAR-dependent LTP, and affective contrast jointly cause reward-memory encoding. I want to learn directed acyclic graphs, do-calculus, and counterfactual inference to move from correlational pharmacological data to mechanistic claims. My hERG cardiotoxicity study, which found that topological features do not beat a plain descriptor baseline (AUROC 0.8426 versus 0.8782), would benefit from causal mediation analysis to understand which molecular properties actually drive toxicity. I meet the technical prerequisites: I work daily with Python, R, and the Linux command line across HPC clusters running Nextflow and SLURM. I am new to formal causal inference methods, having learned Bayesian statistics through PyMC and MCMC on my own. The course fits my career trajectory as I transition from independent research to formal graduate training in digital health, where causal methods will be central to my thesis work on pharmacological mechanism inference. Nigeria and the broader LMIC context shape my research priorities. Addiction neuroscience and antimicrobial resistance are urgent problems in West Africa with limited computational research capacity. Formal training at EMBL-EBI will allow me to bring causal inference methods back to this context, building tools that work with the observational data typical of African clinical settings. SHORT ESSAY: RELEVANCE OF CURRENT WORK My current research involves computational biology, machine learning, and biostatistics across three domains. First, the CCT model for addiction neuroscience uses a system of three coupled ordinary differential equations representing dopaminergic reward prediction error, NMDAR-dependent long-term potentiation, and affective contrast. I calibrated 14 free parameters using Bayesian MCMC with PyMC, drawing priors from a systematic screen of 1,847 records. The model estimates causal effects of pharmacological interventions on reward-memory encoding, directly relevant to the course focus on going beyond associations. Second, my TOPOLOGIX project applies protein-language-model embeddings from ESM-2 combined with Morgan fingerprints and a Random Forest classifier to predict drug-resistance mutations from sequence alone. The model achieves AUROC 0.804 on the Platinum benchmark, covering 100 percent of mutations versus approximately 18 percent for structure-limited tools. This work raises causal questions about which sequence features drive resistance, questions I currently cannot answer formally. Third, my neurocascade simulation engine couples pharmacokinetics to receptor binding to Wilson-Cowan circuit dynamics to behavioral readouts. The model passes 62 of 62 tests and is calibrated with Bayesian methods, but the causal links between receptor-level perturbations and circuit-level behavior remain inferred rather than formally identified. The EMBO course will give me the tools to design experiments and analyses that test these causal claims directly. SHORT ESSAY: LEARNING GOALS AND EXPECTED OUTCOMES I have three specific learning goals for this course. First, I want to master the practical application of directed acyclic graphs and do-calculus to pharmacological data. My CCT model currently uses Bayesian inference to estimate parameters, but I cannot formally test whether the causal structure I assume matches the data. I will learn to construct DAGs for my three-axis system and apply d-separation tests to validate the model topology. Second, I want to learn counterfactual inference methods applicable to observational clinical data. In Nigeria, randomized controlled trials for addiction interventions are rare. I need to estimate what would happen under alternative pharmacological regimens using only observational records. The course tools for counterfactual reasoning will let me answer questions like: would a lower dose of naloxone still prevent reward-memory encoding given the same dopaminergic state? Third, I want to understand how causal representation learning applies to protein-drug interaction data. My TOPOLOGIX model predicts resistance but does not explain why specific mutations cause resistance. I will learn to apply causal discovery algorithms to protein sequence and structure data, moving from prediction to mechanistic explanation. The expected outcome is a formal causal analysis framework I can apply to all three of my active research lines. I will produce a methods note applying DAG-based causal inference to the CCT model within three months of the course, and integrate causal discovery into the TOPOLOGIX pipeline within six months. SHORT ESSAY: DIVERSITY AND CAREER CONTEXT I am a Nigerian researcher working independently without institutional affiliation or PhD supervision. This position is unusual in computational biomedicine, where most researchers operate within well-funded European or North American labs. My research has been self-directed, funded by personal savings and small grants, and conducted on my own computing infrastructure: self-hosted Linux servers running DuckDB-based pipelines and local LLM serving for literature analysis. The LMIC context shapes my research in concrete ways. Addiction neuroscience in Nigeria has almost no computational research capacity. Antimicrobial resistance surveillance relies on genomic pipelines I helped build during my time at GHRU-GSAR, but causal analysis of resistance mechanisms is absent. By attending this course, I will bring formal causal inference methods to a region where they are not taught and rarely applied. I have already built international collaborations with Kent Berridge at Michigan, Samuel Gershman at Harvard, Nathaniel Daw at Princeton, and Marcelo Mattar at NYU. These relationships emerged from my preprints and computational work, not from institutional affiliation. The EMBO course will deepen my methodological skills and expand my network, helping me transition from independent researcher to formal graduate student at Hasso Plattner Institute. EMBO explicitly encourages applications from underrepresented regions. I represent a demographic and geographic group that is severely underrepresented in computational biomedicine: an early-career African researcher working independently in addiction neuroscience and protein ML. My participation will demonstrate that high-quality computational research can originate outside traditional centers, and that causal methods training can have outsized impact when brought to LMIC settings. CHECKLIST - [ ] Complete EMBO online application form at the course website - [ ] Upload this motivation letter as a single PDF - [ ] Upload CV including ORCID, GitHub, publications, and employment history - [ ] Provide contact details for two referees (Berridge, Gershman, or Mattar) - [ ] Confirm availability for the full course dates at EMBL-EBI, Hinxton, UK - [ ] Verify Linux command line and Python/R proficiency meets prerequisites - [ ] Prepare a one-page summary of current research for potential poster session EDITOR NOTES - Eligibility risk: The course targets PhD students and postdocs. Eniola is enrolled in an M.Sc. starting winter 2026/27 but is currently an independent researcher. The application should emphasize his research output (three preprints, one paper under review) to demonstrate he operates at PhD-equivalent level. Verify whether the course accepts pre-PhD researchers with substantial publication records. - Fact verification needed: Confirm the exact AUROC values for TOPOLOGIX on Platinum benchmark (0.804 stated) and SKEMPI 2.0 (0.634 stated). Also confirm the hERG study AUROC values (0.8426 for topological features, 0.8782 for baseline). These numbers come from the applicant profile but should be checked against the actual preprints. - Gap to fill: The application does not specify how Eniola will fund travel and accommodation. The course website should be checked for available fellowships or travel grants for LMIC participants. If none exist, the applicant needs a funding plan before applying.