Publications and Activities

Bioinformatics and AI-assisted drug discovery PhD Student at UCL. Holds extensive experience in protein biochemistry and cancer research. Proficient in developing machine learning models focusing on antibody and TCR-based drug engineering and discovery.

Published Papers

Inhibition of lactate transport by MCT-1 blockade improves chimeric antigen receptor T-cell therapy against B-cell malignancies
From previous work at the UCL Cancer Institute: Small drug molecules for MCT-1 inhibition were used for exploring their impact on CAR T cells, specifically their expansion, IFN-γ production, and interaction with tumor cells. The potential of targeting lactate metabolism via MCT-1 in combination with CAR T cell therapies against B-cell malignancies is explored. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10314680/

Papers under Review / Write-ups

Predicting the Quality of CDR-H3 Antibody Loop Structural Models
Although recent advances in the field have improved antibody structural model prediction accuracy, the results for CDR-H3 are still inconsistent and require further improvement. Providing a confidence score for the structure predictions aids in differentiating well-modelled structures from incorrectly mod- elled structures, giving the user a clearer understanding of the reliability of the 3D-model. We present a 3D-model quality predictor, combining domain knowl- edge with machine learning techniques to predict the accuracy of CDR- H3 3D-models. https://github.com/LilianDenzler/qualiloop_final/blob/acrm/paper.pdf
Accompanying Webserver (currently being migrated, under reconstruction) http://www.bioinf.org.uk/abs/qualiloop/

T-Cell Receptor Numbering Tool http://github.com/LilianDenzler/TCRpaper/blob/main/TCR_Numbering-1.pdf Accompanying Webserver (currently being migrated, under reconstruction) http://www.bioinf.org.uk/abs/qualiloop/TCRnum/TCR_abnum.html

Enhancing Antibody Model Accuracy by Optimizing CDR-H3 Loop Angle Optimize the angle of the CDR-H3 loop in antibody models, enhancing stability and model feasibility. This tool addresses common inaccuracies in loop placement found in models generated by state-of-the-art modelling software, despite having acceptable loop structure accuracy. https://github.com/LilianDenzler/loopflapper

Protein protein interaction benchmark set for machine learning methods development Accurate prediction of protein-protein interface given multiple binder proteins as input. Existing methods for antibody-antigen interface prediction were validated on a small set of antibody-antigen complexes, generalizability is uncertain.

Activities

Bioinformatics and AI-assisted drug discovery PhD Student at UCL. Holds extensive experience in protein biochemistry and cancer research. Proficient in developing machine learning models focusing on antibody and TCR-based drug engineering and discovery.