Preprints & work in press
-
2026
Peer-reviewed articles & book chapters
-
2026Development and validation of an LNA-based multiplex RT-qPCR assay for differentiating Betaarterivirus europensis (PRRSV-1), Betaarterivirus americense (PRRSV-2), and the highly pathogenic L8 lineage of PRRSV-2
The Veterinary Journal 2026;318:106731.
-
2016Systems Vaccinology: Applications, Trends, and Perspectives
In: Vaccine Design. Methods and Protocols, Vol. 1: Vaccines for Human Diseases. Methods Mol Biol 2016;1403:107–30. Springer. Book chapter
-
2015Computational peptide vaccinology
Methods Mol Biol 2015;1268:291–312. Book chapter
-
2012Synthetic lethality for linking the mycophenolate mofetil mode of action with molecular disease and drug profiles
Molecular BioSystems 2012;8(12):3197–207.
-
2011Beyond epitopes: future and application of computational vaccinology
Human Vaccines 2011;7(7):795–7. Commentary
-
2011Mapping of molecular pathways, biomarkers and drug targets for diabetic nephropathy
Proteomics Clinical Applications 2011;5(5–6):354–66.
-
2010Concept and application of a computational vaccinology workflow
Immunome Research 2010;6(Suppl 2):S7.
-
2008Analysis and prediction of protective continuous B-cell epitopes on pathogen proteins
Immunome Research 2008;4:1.
-
2008Mouse mammary tumor virus integration site selection in human and mouse genomes
Journal of Virology 2008;82(3):1360–7.
-
2007Towards a consensus on datasets and evaluation metrics for developing B-cell epitope prediction tools
Journal of Molecular Recognition 2007;20(2):75–82. Community consensus
-
2006Selection and combination of machine learning classifiers for prediction of linear B-cell epitopes on proteins
Journal of Molecular Recognition 2006;19(3):209–14.
-
2006Machine learning approaches for prediction of linear B-cell epitopes on proteins
Journal of Molecular Recognition 2006;19(3):200–8.
-
2003Functional selection of vaccine candidate peptides from Staphylococcus aureus whole-genome expression libraries in vitro
Infection and Immunity 2003;71(8):4633–41.
-
2002Identification of in vivo expressed vaccine candidate antigens from Staphylococcus aureus
Proceedings of the National Academy of Sciences USA 2002;99(10):6573–8.
Patent applications
Three international (PCT) applications as co-inventor, filed by the respective employers. Publication numbers link to the full text on Google Patents.
-
2021Propionibacterium acnes prophylactic and therapeutic immune treatment
WO 2021/165543 A1 · Applicant: Origimm Biotechnology GmbH · Filed 22 Feb 2021, published 26 Aug 2021, earliest priority 21 Feb 2020.
-
2011Critical gene targets for cytotoxic therapy
WO 2011/144738 A1 · Applicant: emergentec biodevelopment GmbH · Filed 20 May 2011, published 24 Nov 2011.
-
2003S. aureus antigene
WO 2004/013166 A2 · Applicant: Intercell Austria AG · Filed 22 Jul 2003, published 12 Feb 2004.
Open-source projects
Methods I have released publicly under MIT, at github.com/sodatana. Published code is the part of the record you can actually run.
-
Pythonmimotope-maker
Peptide mimotopes for discontinuous (conformational) epitopes. It works from an ensemble of structures rather than one static model, identifying surface residues that stay close to each other as the protein moves, and proposes linear peptides mimicking that surface — ranked, and in L, retro-inverso and chimeric L/D forms. The original in-house Perl pipeline is kept alongside the Python implementation.
-
Pythoncdsbo — mRNA coding sequence bicodon optimization
Codon optimization that scores adjacent codon pairs instead of single codons, so codon-context effects survive the optimization. A Viterbi dynamic-programming pass finds the globally optimal coding sequence rather than making a greedy per-residue choice, against bicodon tables derived from human tissue-specific proteomics data — swappable for other expression systems. Kozak context and forbidden patterns such as restriction sites are handled explicitly.
-
PythonLLM-offloader_benchmark
A test suite for deciding which parts of an AI-assisted coding workflow can be handed to a local model and which cannot. It runs identical graded tasks on a local LLM (Ollama / gemma4) and on a frontier model under the same constraints, scores both deterministically against ground truth, and pairs quality with token cost into a per-task offload verdict. The motivation is practical: keep sensitive work on-device, and spend frontier-model budget where it changes the answer.
What the record is about
Four threads run through twenty-plus years of it, and all four are still live in client work:
- Epitopes and the machine learning around them — the PhD work at Intercell (PNAS 2002, Infect Immun 2003) and the methodological papers that followed (J Mol Recognit 2006 ×2, Immunome Res 2008), including the community benchmarking consensus (J Mol Recognit 2007). This is the foundation under today's epitope prediction and antigen selection work.
- Computational vaccinology as a workflow — not a single prediction but a pipeline with defensible decision points: Immunome Res 2010, Hum Vaccin 2011, and the two Methods in Molecular Biology chapters (2015, 2016) that document the method end to end.
- Systems biology, biomarkers and drug repositioning — protein/drug/disease networks applied to therapy selection and target triage: Proteomics Clin Appl 2011, Mol Biosyst 2012, and the cytotoxic-target patent.
- Engineered proteins against hard infections — the current thread: multi-parameter optimization of a lysin into a Fc-fused ribolysin against S. aureus (bioRxiv 2026), and molecular diagnostics for veterinary pathogens (Vet J 2026).
If a prospective project sits near any of these, there is published method behind the advice — and where there is not, I will say so.