FROM DETERMINISTIC CONVERGENCE TO DISCOVERY

FROM DETERMINISTIC CONVERGENCE TO DISCOVERY
Designing the Next Generation of Therapeutics
Matthew L. Hardy

Ninety percent of drug candidates that enter clinical trials fail - and half of those failures happen in Phase II, where the drug was safe but simply didn't work. The standard explanation is "translational failure." Hardy argues that's a description, not a diagnosis: the real cause is architectural. Targets are validated for visibility - expression levels, knockdown effects, biochemical potency - not for structural necessity within the disease system that actually has to be defeated.

Drawing on the PROTEUS cross-domain validation dataset - twenty-six studies spanning cardiovascular, neurological, oncological, and hepatic disease - this book builds the mathematical case that most disease-driving systems are architecturally distributed, not concentrated: redundant, reserve-holding, and built by billions of years of selection against fragility. Hardy introduces two operating tools for telling the difference before a program burns years and capital finding out the hard way:

• The Herfindahl-Hirschman Index, adapted from market-concentration economics, to classify a biological system as distributed, mixed, or concentrated.
• The concordance metric Cd, derived from a Sherrington-Kirkpatrick spin-glass formulation of variant fitness, to quantify whether a specific target is load-bearing, a bridge node, or architecturally redundant - i.e., whether hitting it will produce a durable clinical response or just a very well-characterized molecule with no future.

Chapter One lays the empirical and mathematical foundation; the eleven chapters that follow carry the framework through multi-omic interpretation, structural biology, protein and small-molecule design, synthesis, specialized modalities for CNS and "undruggable" targets, and clinical translation - closing with a quick-reference decision framework that compresses the full pipeline into a single navigable tool.

This is the companion volume to Deterministic Convergence, extending its architectural thesis - that biological systems must be read as architectures, not scored lists of parts - into an applied design discipline for the next generation of therapeutics. Written for computational biologists, medicinal chemists, and translational scientists who want the mathematics that separates a target worth pursuing from a beautiful dead end.