#32 · Systems biology

Path-integral and graph models infer treatment and fitness dynamics from limited data

The common thread is the use of explicit mechanistic or landscape models, calibrated to experimental measurements, to infer unobserved dynamics and to reason about interventions. The antibody affinity-maturation study pairs a stochastic model, calibrated to in vivo lineage and deep mutational scanning data, with a path-integral method that identifies optimal antigen-dosing protocols. The acute myeloid leukemia framework uses an edge-weighted digraph to represent cytarabine treatment contextuality and to explore SAMHD1 inhibitors, delivered as a web-based tool. ALFA-K infers chromosome-level karyotype fitness landscapes from longitudinal single-cell data and predicts emergent aneuploid karyotypes. The shared element is that model structure carries the inference, while the distinct contribution across items is the move from fitting observed dynamics toward evaluating and prescribing intervention outcomes.