Draft — BSPR Annual Scientific Meeting 2026 Bursary

MOTIVATION LETTER

The BSPR Annual Scientific Meeting 2026 is the one venue where my current research program, TOPOLOGIX, belongs in front of a proteomics audience. I am an independent computational researcher with a B.Pharm from the University of Ibadan, currently enrolled in the M.Sc. Digital Health program at the Hasso Plattner Institute / University of Potsdam. My work sits at the intersection of protein language models, topological data analysis, and drug-resistance prediction, and the meeting's emphasis on computational proteomics and data analysis is precisely the community that needs to see my results.

TOPOLOGIX predicts drug-resistance mutations from protein sequence alone. The system uses ESM-2 protein language model delta-embeddings combined with Morgan/ECFP drug fingerprints fed into a Random Forest classifier. On the Platinum benchmark of 553 mutations, TOPOLOGIX achieves an AUROC of 0.804 with a standard deviation of 0.025. On SKEMPI 2.0, it reaches 0.634. These numbers beat structure-based baselines such as mCSM-lig at approximately 0.70, but the critical advantage is coverage: TOPOLOGIX covers 100 percent of mutations, while structure-limited tools cover only about 18 percent because they require resolved crystal structures. This is a computational proteomics result that directly addresses the field's bottleneck: most clinically observed mutations occur in proteins without high-resolution structures.

The path to TOPOLOGIX included a negative result worth stating plainly. My earlier work tested whether bipartite persistent homology, using an opposition-distance metric implemented in Ripser and GUDHI, could predict hERG cardiotoxicity from protein-ligand interface geometry. A pre-registered, powered replication found that topological features did not beat a plain descriptor baseline, with an AUROC of 0.8426 versus 0.8782. I then applied the same topological constructs to drug-resistance prediction and found they carried almost no signal, with AUROCs of 0.425 and 0.485 on the Platinum benchmark. That falsification motivated the pivot to sequence representations, which produced TOPOLOGIX. I report these results because the BSPR community values methodological rigor, and knowing what does not work in protein-ligand interface analysis is as important as knowing what does.

Attending the meeting and the post-conference EMBL-EBI bioinformatics training workshop would directly strengthen my proteomics data analysis skills. The workshop's coverage of PRIDE and QuantMS tools maps onto the gap in my current pipeline: I work with sequence and mutation data, but I have not yet integrated mass spectrometry proteomics evidence into my resistance-prediction framework. The M.Sc. Digital Health program at HPI has given me a foundation in health data infrastructure, and the BSPR workshop would add the proteomics-specific data handling layer that my current work lacks.

I am an independent researcher without institutional laboratory funding. My prior work has been self-supported, with preprints deposited on OSF and Zenodo and currently under review at peer-reviewed journals including IART, PNPBP, and NBR. A bursary would make attendance feasible and would connect my independent computational work to the established proteomics research community in the UK and Europe.

RESEARCH STATEMENT

My research program addresses a specific problem in computational proteomics: predicting which mutations in a protein confer drug resistance, using only sequence information, when structural data is unavailable for most clinically relevant targets. The TOPOLOGIX project is the current core of this program, and it emerged directly from a falsified hypothesis about protein-ligand interface geometry.

The negative result came first. I tested whether bipartite persistent homology, a topological data analysis method using an opposition-distance metric implemented in Ripser and GUDHI, could predict hERG cardiotoxicity from protein-ligand interface geometry. The study was pre-registered and powered. The result was unambiguous: topological features achieved an AUROC of 0.8426, while a plain descriptor baseline achieved 0.8782. The topological features did not add predictive value. I then applied the same class of topological constructs to drug-resistance prediction, testing whether interface geometry drives resistance. On the Platinum benchmark, the topological approach achieved AUROCs of 0.425 and 0.485, essentially no signal. These two results, both reported directly rather than reframed, ruled out interface geometry as the driver of resistance and motivated a sequence-representation approach.

TOPOLOGIX is that sequence approach. The system uses ESM-2 protein language model delta-embeddings to represent the mutation's effect on the protein's learned sequence context, combines these with Morgan/ECFP drug fingerprints, and classifies resistance with a Random Forest. On the Platinum benchmark of 553 mutations, TOPOLOGIX achieves an AUROC of 0.804 with a standard deviation of 0.025. On SKEMPI 2.0, it achieves 0.634. The comparison to structure-based tools is the key result: mCSM-lig achieves approximately 0.70 AUROC, but it can only score mutations in proteins with resolved structures, covering roughly 18 percent of the Platinum benchmark. TOPOLOGIX covers 100 percent of mutations. For clinical and surveillance applications, where the resistance mutation is often in a protein that has never been crystallized, coverage is the binding constraint.

The methods are reproducible. The pipeline is implemented in Python using scikit-learn for the Random Forest, Hugging Face transformers for ESM-2 embedding extraction, and RDKit for Morgan fingerprint generation. The code is available on GitHub under my account, AmunRaPtah. The pre-registration and benchmark evaluation protocols are documented on OSF.

The BSPR Annual Scientific Meeting's focus on proteomics technology and data analysis is the correct venue for this work because TOPOLOGIX is fundamentally a proteomics informatics result. It answers a question that structural biologists and proteomics researchers face daily: what can we predict about a protein's function and drug interactions when we only have its sequence? The meeting's training workshop on EMBL-EBI bioinformatics resources, including PRIDE and QuantMS, would extend my pipeline to incorporate mass spectrometry evidence. My current framework uses sequence and mutation data only. Integrating proteomics evidence would allow TOPOLOGIX to validate predicted resistance mutations against observed protein abundance and modification data, strengthening the biological grounding of the predictions.

My broader research program includes the CCT model, a tripartite pharmacological framework for reward-memory encoding prevention in addiction, which uses coupled ODE systems calibrated with Bayesian MCMC against literature-elicited priors from an 1,847-record screen. That work is under review at three journals. I also maintain neurocascade, a receptor-to-behavior brain-circuit simulation engine with 62 passing tests. These projects share a methodological commitment with TOPOLOGIX: build explicit models, pre-register the hypotheses, report the negative results, and let the data decide. The BSPR meeting is where computational proteomics methods are discussed, and my sequence-based resistance prediction work belongs in that conversation.

ESSAY: RELEVANCE TO PROTEOMICS AND COMPUTATIONAL BIOLOGY

The BSPR Annual Scientific Meeting centers on proteomics technology, data analysis, and computational approaches. My TOPOLOGIX project is a computational proteomics application that predicts drug-resistance mutations from protein sequence using protein language model embeddings. The relevance to the meeting is direct: TOPOLOGIX addresses the challenge of extracting functional predictions from protein sequence data at scale, which is a core concern of computational proteomics.

The meeting's emphasis on data analysis matches my methodological profile. TOPOLOGIX uses ESM-2 delta-embeddings, a protein language model that captures evolutionary and structural information from sequence alone. The system combines these embeddings with Morgan/ECFP drug fingerprints and classifies with a Random Forest. This is a modern machine learning pipeline applied to a proteomics problem, and the BSPR audience includes researchers developing and evaluating exactly these kinds of computational tools.

The post-conference EMBL-EBI bioinformatics training workshop is directly relevant to my current enrollment in the M.Sc. Digital Health program at the Hasso Plattner Institute. The workshop's coverage of PRIDE and QuantMS would add mass spectrometry proteomics data handling to my skill set, which currently spans Python, PyMC for Bayesian calibration, topological data analysis with Ripser and GUDHI, and molecular dynamics with GROMACS. The workshop would close a specific gap: my resistance-prediction pipeline does not yet incorporate proteomics evidence, and the training would enable that integration.

My prior work demonstrates the rigor that the BSPR community expects. The hERG cardiotoxicity topology study was pre-registered and powered, and it produced a negative result that I reported directly. The interface-topology-for-resistance study similarly ruled out a hypothesis. These results are published as preprints on OSF and Zenodo and are under review at peer-reviewed journals. TOPOLOGIX, the positive result that emerged from those falsifications, achieves an AUROC of 0.804 on the Platinum benchmark with 100 percent mutation coverage. This is a concrete, reproducible computational proteomics contribution.

ESSAY: POTENTIAL BENEFIT FROM ATTENDANCE AND TRAINING

Attendance at the BSPR Annual Scientific Meeting 2026 would benefit my research in three specific ways. First, the meeting would expose TOPOLOGIX to the proteomics research community, including researchers who work on mass spectrometry data, protein structure prediction, and computational methods for protein function annotation. My work currently circulates in preprint form on OSF and Zenodo, and the meeting is an opportunity to present it to a targeted audience and receive direct feedback.

Second, the post-conference EMBL-EBI bioinformatics training workshop would add a concrete skill to my pipeline. The workshop covers PRIDE and QuantMS, tools for proteomics data deposition and quantification. My current TOPOLOGIX pipeline uses sequence and mutation data only. Integrating mass spectrometry proteomics evidence would allow me to validate predicted resistance mutations against observed protein abundance data, which would strengthen the biological grounding of the predictions. The M.Sc. Digital Health program at HPI has trained me in health data infrastructure, and the BSPR workshop would add the proteomics-specific data layer.

Third, the meeting would connect me to the broader computational proteomics community. I am an independent researcher based in Nigeria, currently enrolled in a German master's program. I do not have institutional laboratory funding, and my work is self-supported. The BSPR meeting and workshop represent an opportunity to build collaborations with researchers who have access to proteomics experimental data, which my computational pipeline currently lacks. A bursary would make this attendance feasible.

CHECKLIST

- [ ] Confirm BSPR membership status and join if required for bursary eligibility
- [ ] Verify bursary application deadline on BSPR website
- [ ] Confirm whether abstract submission is required for bursary consideration
- [ ] Prepare CV in BSPR format, emphasizing TOPOLOGIX and computational proteomics methods
- [ ] Obtain letter of support from a BSPR member or academic collaborator
- [ ] Verify registration fee amount and bursary coverage details
- [ ] Confirm EMBL-EBI workshop registration is included or separate
- [ ] Prepare TOPOLOGIX results summary with AUROC figures and benchmark details
- [ ] Confirm ORCID and GitHub links are active and current
- [ ] Draft budget statement for travel and accommodation if required

EDITOR NOTES

- Eligibility risk: BSPR bursaries typically prioritize members and abstract presenters. The 2026 abstract submission window is closed per the strategy notes. The applicant must confirm whether a bursary can be awarded without an accepted abstract, or whether late-breaking submission is possible.
- The applicant's primary research line for this application is TOPOLOGIX, the sequence-based drug-resistance prediction project. The hERG and interface-topology studies are presented as falsified prior work that motivated the pivot, not as current ongoing research. Do not present them as active projects.
- The M.Sc. Digital Health enrollment at HPI/Potsdam is listed as starting Winter Semester 2026/27, while employment at Synthcare is listed as March 2026 to present. Verify the timeline coherence before submission, as the bursary committee may question concurrent full-time employment and full-time study.
- The applicant should insert a specific statement about financial need, including estimated travel, accommodation, and registration costs from Nigeria or Germany, as the bursary criteria require demonstrated lack of alternative funding.

MOTIVATION LETTER

The BSPR Annual Scientific Meeting 2026 is the one venue where my current research program, TOPOLOGIX, belongs in front of a proteomics audience. I am an independent computational researcher with a B.Pharm from the University of Ibadan, currently enrolled in the M.Sc. Digital Health program at the Hasso Plattner Institute / University of Potsdam. My work sits at the intersection of protein language models, topological data analysis, and drug-resistance prediction, and the meeting's emphasis on computational proteomics and data analysis is precisely the community that needs to see my results.

TOPOLOGIX predicts drug-resistance mutations from protein sequence alone. The system uses ESM-2 protein language model delta-embeddings combined with Morgan/ECFP drug fingerprints fed into a Random Forest classifier. On the Platinum benchmark of 553 mutations, TOPOLOGIX achieves an AUROC of 0.804 with a standard deviation of 0.025. On SKEMPI 2.0, it reaches 0.634. These numbers beat structure-based baselines such as mCSM-lig at approximately 0.70, but the critical advantage is coverage: TOPOLOGIX covers 100 percent of mutations, while structure-limited tools cover only about 18 percent because they require resolved crystal structures. This is a computational proteomics result that directly addresses the field's bottleneck: most clinically observed mutations occur in proteins without high-resolution structures.

The path to TOPOLOGIX included a negative result worth stating plainly. My earlier work tested whether bipartite persistent homology, using an opposition-distance metric implemented in Ripser and GUDHI, could predict hERG cardiotoxicity from protein-ligand interface geometry. A pre-registered, powered replication found that topological features did not beat a plain descriptor baseline, with an AUROC of 0.8426 versus 0.8782. I then applied the same topological constructs to drug-resistance prediction and found they carried almost no signal, with AUROCs of 0.425 and 0.485 on the Platinum benchmark. That falsification motivated the pivot to sequence representations, which produced TOPOLOGIX. I report these results because the BSPR community values methodological rigor, and knowing what does not work in protein-ligand interface analysis is as important as knowing what does.

Attending the meeting and the post-conference EMBL-EBI bioinformatics training workshop would directly strengthen my proteomics data analysis skills. The workshop's coverage of PRIDE and QuantMS tools maps onto the gap in my current pipeline: I work with sequence and mutation data, but I have not yet integrated mass spectrometry proteomics evidence into my resistance-prediction framework. The M.Sc. Digital Health program at HPI has given me a foundation in health data infrastructure, and the BSPR workshop would add the proteomics-specific data handling layer that my current work lacks.

I am an independent researcher without institutional laboratory funding. My prior work has been self-supported, with preprints deposited on OSF and Zenodo and currently under review at peer-reviewed journals including IART, PNPBP, and NBR. A bursary would make attendance feasible and would connect my independent computational work to the established proteomics research community in the UK and Europe.

RESEARCH STATEMENT

My research program addresses a specific problem in computational proteomics: predicting which mutations in a protein confer drug resistance, using only sequence information, when structural data is unavailable for most clinically relevant targets. The TOPOLOGIX project is the current core of this program, and it emerged directly from a falsified hypothesis about protein-ligand interface geometry.

The negative result came first. I tested whether bipartite persistent homology, a topological data analysis method using an opposition-distance metric implemented in Ripser and GUDHI, could predict hERG cardiotoxicity from protein-ligand interface geometry. The study was pre-registered and powered. The result was unambiguous: topological features achieved an AUROC of 0.8426, while a plain descriptor baseline achieved 0.8782. The topological features did not add predictive value. I then applied the same class of topological constructs to drug-resistance prediction, testing whether interface geometry drives resistance. On the Platinum benchmark, the topological approach achieved AUROCs of 0.425 and 0.485, essentially no signal. These two results, both reported directly rather than reframed, ruled out interface geometry as the driver of resistance and motivated a sequence-representation approach.

TOPOLOGIX is that sequence approach. The system uses ESM-2 protein language model delta-embeddings to represent the mutation's effect on the protein's learned sequence context, combines these with Morgan/ECFP drug fingerprints, and classifies resistance with a Random Forest. On the Platinum benchmark of 553 mutations, TOPOLOGIX achieves an AUROC of 0.804 with a standard deviation of 0.025. On SKEMPI 2.0, it achieves 0.634. The comparison to structure-based tools is the key result: mCSM-lig achieves approximately 0.70 AUROC, but it can only score mutations in proteins with resolved structures, covering roughly 18 percent of the Platinum benchmark. TOPOLOGIX covers 100 percent of mutations. For clinical and surveillance applications, where the resistance mutation is often in a protein that has never been crystallized, coverage is the binding constraint.

The methods are reproducible. The pipeline is implemented in Python using scikit-learn for the Random Forest, Hugging Face transformers for ESM-2 embedding extraction, and RDKit for Morgan fingerprint generation. The code is available on GitHub under my account, AmunRaPtah. The pre-registration and benchmark evaluation protocols are documented on OSF.

The BSPR Annual Scientific Meeting's focus on proteomics technology and data analysis is the correct venue for this work because TOPOLOGIX is fundamentally a proteomics informatics result. It answers a question that structural biologists and proteomics researchers face daily: what can we predict about a protein's function and drug interactions when we only have its sequence? The meeting's training workshop on EMBL-EBI bioinformatics resources, including PRIDE and QuantMS, would extend my pipeline to incorporate mass spectrometry evidence. My current framework uses sequence and mutation data only. Integrating proteomics evidence would allow TOPOLOGIX to validate predicted resistance mutations against observed protein abundance and modification data, strengthening the biological grounding of the predictions.

My broader research program includes the CCT model, a tripartite pharmacological framework for reward-memory encoding prevention in addiction, which uses coupled ODE systems calibrated with Bayesian MCMC against literature-elicited priors from an 1,847-record screen. That work is under review at three journals. I also maintain neurocascade, a receptor-to-behavior brain-circuit simulation engine with 62 passing tests. These projects share a methodological commitment with TOPOLOGIX: build explicit models, pre-register the hypotheses, report the negative results, and let the data decide. The BSPR meeting is where computational proteomics methods are discussed, and my sequence-based resistance prediction work belongs in that conversation.

ESSAY: RELEVANCE TO PROTEOMICS AND COMPUTATIONAL BIOLOGY

The BSPR Annual Scientific Meeting centers on proteomics technology, data analysis, and computational approaches. My TOPOLOGIX project is a computational proteomics application that predicts drug-resistance mutations from protein sequence using protein language model embeddings. The relevance to the meeting is direct: TOPOLOGIX addresses the challenge of extracting functional predictions from protein sequence data at scale, which is a core concern of computational proteomics.

The meeting's emphasis on data analysis matches my methodological profile. TOPOLOGIX uses ESM-2 delta-embeddings, a protein language model that captures evolutionary and structural information from sequence alone. The system combines these embeddings with Morgan/ECFP drug fingerprints and classifies with a Random Forest. This is a modern machine learning pipeline applied to a proteomics problem, and the BSPR audience includes researchers developing and evaluating exactly these kinds of computational tools.

The post-conference EMBL-EBI bioinformatics training workshop is directly relevant to my current enrollment in the M.Sc. Digital Health program at the Hasso Plattner Institute. The workshop's coverage of PRIDE and QuantMS would add mass spectrometry proteomics data handling to my skill set, which currently spans Python, PyMC for Bayesian calibration, topological data analysis with Ripser and GUDHI, and molecular dynamics with GROMACS. The workshop would close a specific gap: my resistance-prediction pipeline does not yet incorporate proteomics evidence, and the training would enable that integration.

My prior work demonstrates the rigor that the BSPR community expects. The hERG cardiotoxicity topology study was pre-registered and powered, and it produced a negative result that I reported directly. The interface-topology-for-resistance study similarly ruled out a hypothesis. These results are published as preprints on OSF and Zenodo and are under review at peer-reviewed journals. TOPOLOGIX, the positive result that emerged from those falsifications, achieves an AUROC of 0.804 on the Platinum benchmark with 100 percent mutation coverage. This is a concrete, reproducible computational proteomics contribution.

ESSAY: POTENTIAL BENEFIT FROM ATTENDANCE AND TRAINING

Attendance at the BSPR Annual Scientific Meeting 2026 would benefit my research in three specific ways. First, the meeting would expose TOPOLOGIX to the proteomics research community, including researchers who work on mass spectrometry data, protein structure prediction, and computational methods for protein function annotation. My work currently circulates in preprint form on OSF and Zenodo, and the meeting is an opportunity to present it to a targeted audience and receive direct feedback.

Second, the post-conference EMBL-EBI bioinformatics training workshop would add a concrete skill to my pipeline. The workshop covers PRIDE and QuantMS, tools for proteomics data deposition and quantification. My current TOPOLOGIX pipeline uses sequence and mutation data only. Integrating mass spectrometry proteomics evidence would allow me to validate predicted resistance mutations against observed protein abundance data, which would strengthen the biological grounding of the predictions. The M.Sc. Digital Health program at HPI has trained me in health data infrastructure, and the BSPR workshop would add the proteomics-specific data layer.

Third, the meeting would connect me to the broader computational proteomics community. I am an independent researcher based in Nigeria, currently enrolled in a German master's program. I do not have institutional laboratory funding, and my work is self-supported. The BSPR meeting and workshop represent an opportunity to build collaborations with researchers who have access to proteomics experimental data, which my computational pipeline currently lacks. A bursary would make this attendance feasible.

CHECKLIST

- [ ] Confirm BSPR membership status and join if required for bursary eligibility
- [ ] Verify bursary application deadline on BSPR website
- [ ] Confirm whether abstract submission is required for bursary consideration
- [ ] Prepare CV in BSPR format, emphasizing TOPOLOGIX and computational proteomics methods
- [ ] Obtain letter of support from a BSPR member or academic collaborator
- [ ] Verify registration fee amount and bursary coverage details
- [ ] Confirm EMBL-EBI workshop registration is included or separate
- [ ] Prepare TOPOLOGIX results summary with AUROC figures and benchmark details
- [ ] Confirm ORCID and GitHub links are active and current
- [ ] Draft budget statement for travel and accommodation if required

EDITOR NOTES

- Eligibility risk: BSPR bursaries typically prioritize members and abstract presenters. The 2026 abstract submission window is closed per the strategy notes. The applicant must confirm whether a bursary can be awarded without an accepted abstract, or whether late-breaking submission is possible.
- The applicant's primary research line for this application is TOPOLOGIX, the sequence-based drug-resistance prediction project. The hERG and interface-topology studies are presented as falsified prior work that motivated the pivot, not as current ongoing research. Do not present them as active projects.
- The M.Sc. Digital Health enrollment at HPI/Potsdam is listed as starting Winter Semester 2026/27, while employment at Synthcare is listed as March 2026 to present. Verify the timeline coherence before submission, as the bursary committee may question concurrent full-time employment and full-time study.
- The applicant should insert a specific statement about financial need, including estimated travel, accommodation, and registration costs from Nigeria or Germany, as the bursary criteria require demonstrated lack of alternative funding.

AI Draft — BSPR Annual Scientific Meeting 2026 Bursary