Amphoraxe Co-Founder and Chief Scientific Officer Dr. Inanç Birol was an invited speaker at the 13th International Meeting on Antimicrobial Peptides (IMAP 2026), held September 2–4 at the Faculty of Science, University of Split, Croatia. IMAP brings together researchers from around the world working on the design, mode of action, and therapeutic applications of antimicrobial peptides (AMPs).
Dr. Birol opened the meeting’s second day with his talk, “From Sequence to Target Hypothesis: Systematic Docking of AMPs Against Essential Bacterial Cell Surface Proteins,” and joined the closing panel discussion on the future of AMPs and AMP research.
Amphoraxe research intern Berke Uçar also attended, presenting a poster on “Pathogen-specific antimicrobial activity prediction with biological large language model-based methods” during the opening day’s poster session. The work introduces triAMPh, a framework that combines graph-based machine learning with features from biological large language models to predict which pathogens a peptide is likely to act against — helping prioritize the most promising candidates before costly wet-lab testing. The poster was co-authored with Amphoraxe’s Dr. Anat Yanai and Dr. Birol, alongside collaborators at BC Cancer, the University of British Columbia, and the University of Victoria.
View the IMAP 2026 Programme & Abstracts →
Moments from IMAP 2026
The poster: triAMPh
Berke Uçar’s poster is reproduced below in two parts. Against AMPSpecies-Specific, an existing species-specific predictor, triAMPh lifted accuracy from 61.1% to 78.7% on a dataset mined from public sources. It was also tested on pathogens it had never seen during training — the case that matters most when screening peptides against a species no model has encountered before.
Click the arrows to move between the two parts, or open the PDF below to read the figures and tables at full size.