I am Tharuniya Anpalakan, a Biomedical Science Undergraduate with research interests across molecular biology, bioinformatics, computational biology, cancer genomics, drug discovery, systems biology, biomedical artificial intelligence and regenerative medicine. My work brings together biomedical knowledge, original research documentation, public molecular datasets and reproducible computational approaches to investigate biologically meaningful questions.
My research portfolio includes co-authored scholarly publications, a documented experimental microbiology investigation, and computational studies spanning tumour transcriptomics, single-cell biology, drug–target interactions, antimicrobial-resistance genomics, organoid transplantation, molecular pathways and evidence reliability in biomedical AI.
I approach research by connecting biological mechanisms with transparent methodology, quantitative analysis, literature-supported interpretation and clearly defined limitations. I am especially interested in work that strengthens the link between laboratory science, molecular evidence and translational research.
My research documentation includes academic contexts associated with Northumbria University and the University of Sri Jayewardenepura. Each linked project distinguishes original experimental contributions, collaborative scholarly work and secondary computational research as appropriate.
Research focus: Biological insight · Evidence-based analysis · Reproducible workflows · Clear scientific communication
I have contributed to a published co-authored journal review and a first-author conference abstract, alongside developing independent computational research portfolios. My ResearchGate profile provides an additional research presence. The following records are separated by publication type so that a journal article, a conference abstract and an unpublished project report are not confused.
|
Future direction towards human in-vivo organoid transplantation Authors: Neranja Sandamini, Thisuri Cyara Jayaweera, Suha Ahmed Naseerdeen, Tharuniya Anpalakan, Aafrin Aysha Mohammed Safrin. Venue: BMS The Journal of Applied Learning, 3(2), 80–87 (July 2025). Online ISSN 2989-0705. Research area: regenerative medicine, stem-cell-derived organoids, transplantation applications and translational challenges. Contribution status: Co-author (fourth in the printed author order). The existing records do not establish individual author contributions; no sole-authorship claim is made.
|
Evidence of Foodborne Pathogen Contamination in Riverine Fish from Kelani River, Sri Lanka: A Public Health Perspective Authors: Anpalakan T (Tharuniya Anpalakan — first author), Meddage AKMMK, Liyanage GY, Idroos FS and Manage PM. Venue: 12th International Conference on Multidisciplinary Approaches (iCMA 2026), Faculty of Graduate Studies, University of Sri Jayewardenepura. Abstract: PP–137, printed page 189, in the Proceedings of the 12th International Conference on Multidisciplinary Approaches (iCMA 2026), ISSN 2386-1509. Conference dates: 6–7 April 2026. Research area: original riverine fish microbiological surveillance and public health. A subsequent Python-based computational reanalysis is documented separately in the associated repository. Publication evidence: Verified against the author-supplied official conference proceedings PDF, abstract PP–137 (printed p. 189; PDF page 208). The conference's download link is currently unreliable, so I link the research citation and ResearchGate profile instead. A separate DOI and the abstract's peer-review status are not verified.
|
My repositories also include IEEE-style manuscripts, LaTeX reports, computational research write-ups, figures, notebooks and literature/evidence syntheses. These are research communication outputs and should not be presented as peer-reviewed publications without a separately verified venue record.
| Research area | Supporting repository | Type |
|---|---|---|
| Cancer survival analysis | HECKTOR prognosis | Research manuscript and project report |
| Diabetes single-cell pathways | Pathway-aware single-cell study | Completed comparative analysis with independent-cohort validation; research report and validation report |
| Biomedical language-model evidence | Biomedical RAG | Computational report and research code |
| Organoid transplantation evidence mapping | Organoid landscape | Independent computational report; separate from the 2025 journal review |
| Microbial surveillance reanalysis | Kelani River project | Computational follow-up report; separate from the 2026 conference abstract |
| Polyherbal antidiabetic and antioxidant analysis | Experimental record and computational reanalysis | Completed public-release analysis with a research report |
| Drug targets and molecular networks | Antiepileptic drugs · Autoimmune biologics | Portfolio research reports |
There are 13 research repositories plus this profile repository. The 12 projects featured below are grouped with project-specific methods, results, limitations and direct links; the additional completed polyherbal reanalysis is linked in the research-outputs table above.
|
Cancer Genomics • Microarray • Molecular Pathways Computational reanalysis of GEO GSE45827 comparing source-labelled Luminal A and Luminal B tumours using batch-adjusted differential expression and Reactome enrichment. Research highlights
Methods & tools: Python · GEO · Differential Expression · Reactome · PCA Evidence boundary: The results are associations in one public microarray cohort, not a diagnostic or prognostic test. 
|
Experimental Microbiology • Public Health • Statistics Documents an original fish microbiology investigation and a separate computational reanalysis of published aggregate findings from Kelani River, Sri Lanka. Research highlights
Methods & tools: Microbiology · Python · Public Health · Reanalysis · Research Reporting Evidence boundary: The computational extension adds no specimens; aggregate records do not independently establish every reported p-value.
|
|
Biomedical AI • LLM Evaluation • Retrieval Controlled evaluation of how gold, missing, irrelevant, topic-matched and contradictory evidence affect a biomedical question-answering model. Research highlights
Methods & tools: Python · PubMedQA · Qwen2.5 · BM25 · RAG · Evaluation Evidence boundary: Results describe this experimental setup; confidence did not reliably flag contradictory evidence.
|
Computational Pharmacology • Knowledge Networks Evidence-curated comparison of antiseizure medications through molecular targets, drug chemistry, mechanism classes, selected pharmacokinetic summaries and networks. Research highlights
Methods & tools: RDKit · NetworkX · PubChem · ChEMBL · UniProt Evidence boundary: Descriptive research portfolio, not a prescribing or treatment recommendation system.
|
|
Cancer Biology • Biomarkers • Public Gene Expression Combines a clearly marked educational clinical-case reanalysis with a separate public GSE54129 gastric expression analysis. Research highlights
Methods & tools: Python · GEO · Biomarker Analysis · Statistical Testing Evidence boundary: The educational case and public cohort are separate; results are not clinical diagnosis.
|
Natural Products • Molecular Docking • Evidence Appraisal Completed computational study of literature-qualified polyherbal candidates using receptor validation, structural preparation and exploratory docking with explicitly defined geometric PASS criteria. Research highlights
Methods & tools: AutoDock Vina · Python · Phytochemistry · Molecular Modelling Evidence boundary: This manuscript is not peer-reviewed. A docking PASS denotes geometric pose consistency; docking does not establish experimental binding affinity, enzyme inhibition or clinical efficacy.
|
|
Systems Biology • Oxidative Stress • Molecular Networks Maps a curated redox and vascular-biology gene set onto STRING functional associations and Reactome pathway annotations. Research highlights
Methods & tools: Python · STRING · UniProt · Reactome · NetworkX Evidence boundary: A knowledge-guided network, not evidence of direct protein binding or causal pathway activation.
|
Bacterial Genomics • AMR • Reproducible QC Completed, reproducible acquisition and assembly-QC workflow for four NCBI RefSeq Staphylococcus aureus assemblies, with checksum validation, automated tests, executed notebooks and an updated research report. Research highlights
Methods & tools: NCBI RefSeq · FASTA · SHA-256 · Python · Jupyter · Assembly QC Evidence boundary: AMRFinderPlus, virulence analysis, MLST and SCCmec typing remain unperformed; no antimicrobial-resistance genes, virulence determinants, susceptibility phenotypes or sequence types are claimed.
|
|
Immunology • Biologics • Precision Medicine Evidence-audited analysis connecting biologic therapies with human molecular targets, immune signalling, disease indications and selected clinical studies. Research highlights
Methods & tools: Python · NetworkX · UniProt · Clinical Evidence Curation Evidence boundary: Approval status is indication-, jurisdiction- and date-specific; network structure does not rank clinical efficacy.
|
Regenerative Medicine • Organoids • Translational Biology Evidence mapping of preclinical organoid graft studies, tissue sources, experimental models, graft integration and translational limitations. Research highlights
Methods & tools: Evidence Synthesis · NetworkX · PubMed · Translational Research Evidence boundary: Bounded preclinical evidence map, not an exhaustive systematic review or clinical success estimate.
|
|
Single-Cell RNA-seq • Donor-Level Pseudobulk • KEGG Pathways Completed donor-level analysis of human pancreatic beta-cell transcriptomes, integrating covariate-adjusted differential expression, KEGG enrichment and controlled gene-, pathway- and matched-random modelling. Research highlights
Methods & tools: Python · Donor-Level Pseudobulk · Differential Expression · KEGG · scikit-learn Evidence boundary: External estimates are exploratory because the independent cohort contains only ten donors; the results do not establish clinical predictive performance.
|
Survival Analysis • Clinical Modelling • Validation Develops a survival-analysis pipeline with Kaplan–Meier, Cox regression, cross-validation and synthetic demonstrations of centre-wise assessment. Research highlights
Methods & tools: Python · Survival Analysis · Cox PH · Cross-Validation Evidence boundary: The real dataset is apparently single-centre; synthetic centre splits are not true external validation.
|
|
|
|
|
|
|
Learning note: These are tools featured across my projects, not an expert-level proficiency list. I continue to learn them through guided practice, experimentation, documentation and revision.
The toolkit reflects methods and platforms used or documented across linked projects; implementation details and completed analyses are described in each repository.
Biomedical question and biological context
↓
Literature and evidence appraisal
↓
Study design and dataset audit
↓
Source provenance and quality checks
↓
Reproducible computational workflow
↓
Statistical and biological interpretation
↓
Robustness checks and known limitations
↓
Figures, notebooks and scientific report
↓
Open, traceable research outputs
Ask precisely → check the evidence → analyse transparently → report responsibly.
| Use traceable literature, datasets and identifiers. | Distinguish observation, inference and hypothesis. | Report quantitative outcomes with their limits. | Preserve scripts, notebooks and research provenance. |
- Centre-aware clinical prognosis: pursue genuine external validation when suitable multi-centre data become available.
- Single-cell diabetes research: extend the completed donor-level benchmark to larger independent cohorts and alternative pathway-scoring methods.
- MRSA genomics: execute and review AMRFinderPlus screening with documented versions and database coverage.
- Natural products and docking: continue staged structural validation while respecting source chemistry and assay limitations.
- Biomedical AI reliability: investigate evidence sensitivity, miscalibration and practical detection of unsupported answers.
- Translational evidence mapping: maintain a clear boundary between preclinical observations and clinical applicability.
These are research directions described or motivated by my existing projects—not a claim that the listed next steps have already been completed.
I welcome conversations around biomedical research, bioinformatics, experimental microbiology, molecular mechanisms, computational pharmacology, reproducible analysis and responsible biomedical AI.
