Computational modeling of platelet activation signatures in response to diverse immune and hemostatic agonists

Platelets are increasingly recognized as key players not only in hemostasis, but also in immunity and inflammation. However, the mechanisms and markers underlying their activation remain incompletely understood. This study aimed to decipher how platelets respond to different stimuli and to identify specific molecular signatures using computational approaches. Platelets from 10 healthy donors were stimulated under seven conditions, including TRAP (PAR-1), AYPGKF (PAR-4), ADP, collagen, sCD40L, fibrinogen, and a control. A total of 47 markers-encompassing membrane proteins, soluble mediators, and intracellular signals-were analyzed. Statistical and machine learning methods, including hierarchical clustering and random forest algorithms, were used to classify and interpret the data. Distinct activation profiles emerged for each agonist. A reduced panel of six markers (AKT, CD40L, CD62P, PKC, RANTES, and TSLP) enabled identification of the stimulus with 86.8% accuracy. Machine learning further improved classification (87.9% multiclass accuracy). Differences were also observed across donors, highlighting inter-individual variability. This work supports a new paradigm in which platelets act as "biological sensors," fine-tuning their responses to environmental cues. The identified biomarker panel provides a basis for further investigation into the characterization of platelet activation profiles, with potential relevance for future diagnostic and therapeutic applications in thromboinflammatory and immune-mediated conditions.