Artificial Intelligence Platform for Molecular Targeted Therapy

ARIEL FERNANDEZ

World Scientific Publishing

Lugar: Singapur; Año: 2021 p. 450

9789811232305

This book approaches drug design and, more broadly, molecular targeted therapy, through the translation of fundamental advances in biophysics and physical chemistry. The translation is enabled and leveraged by artificial intelligence (AI), which becomes instrumental as complex biological contexts need to be modeled or recreated. Chapter 1 introduces AI in a heuristic fashion, delineating the basic architectures, operative work flows and free open-source software libraries for deep learning systems and their underlying convolutional neural networks (CNNs). Chapter 2 introduces interfacial physics as required to provide a representational framework to be integrated in the software library of a CNN. The chapter lays the groundwork for AI-leveraged engineering of drug-target interfaces. Chapter 3 presents a first illustration of the power of AI to handle biological complexity at molecular scales: Through AI, an in vivo context capable of expediting the protein folding process is recreated. Thus, in vivo folding pathways are generated and validated vis-à-vis experimental evidence on the converging state of the chain. Chapter 4 explores functional genomics databases from the vantage point of epistructural biology, the conceptual framework introduced in Chapter 2 to enable the implementation of AI methods. Equipped with the conceptual and representational arsenal described, the chapter delineates an AI platform to infer amyloidogenic propensity at a genome-wide level. Chapter 5 introduces an evolutionary axis to epistructural biology, needed to develop drug selectivity filters capable of telling apart paralogous proteins and avoiding off-target paralogs. The chapter emphasizes that the protein-water interface constitutes a unique marker for molecular evolution. This fact endows epistructural biology with a vantage point to explore the evolutionary axis, which is essential to understand and manipulate biological phenomena. Chapter 6 delineates biological function from the perspective of epistructural biology and assesses the functional impact of molecular targeted therapy guided by AI-empowered epistructural design. Chapter 7 illustrates the power of epistructure-based selectivity filters in AI-assisted drug design. Chapter 8 introduces molecular targeted therapies that get redesigned to effectively synergize with cancer immunotherapies. The empowerment of the latter is guided by epistructural AI-leveraged design. Chapter 9 deals with choreographed synergies between targeted therapy and immunotherapy guided by AI. The goal is to leverage cancer evolution as it evades the elicited immune response, thereby steering the tumor towards its annihilation with the aid of AI. Chapter 10 deals with the possibility of leveraging AI to empower molecular dynamics (MD), whose true potential remains largely untested because relevant timescales are often beyond reach and infrequent events of direct relevance to biophysical processes are missed. We apply deep learning to a) encode the dynamics into a simplified embodiment that retains only essential topological features of the vector field that steers MD integration, b) propagate the simplified trajectory beyond the timescales accessible to atomistic MD, and c) reconstruct the coarse grained trajectory back at the atomistic level. Thus, AI constructs a generative model that provides hierarchical understanding of MD simulation while enabling access to realistic timescales and the capture of physically meaningful rare events. Chapter 11 deals with rational drug design in the absence of reported structure for the therapeutic target and information on its regulation. This chapter delineates the leveraging of AI to steer the engineering of molecular targeted therapy in such challenging situations. The chapter focuses specifically on protein associations, where drug-based disruption of dysfunctional complexes poses a major challenge to targeted therapy. The problem becomes daunting when the structure and regulated modulation of the complex are unknown. To address the challenge, we leverage an AI platform that learns from epistructural information and infers regulation-susceptible regions that also generate interfacial tension between protein and water, thereby promoting contextually tuned protein associations. The AI-empowered discovery platform is benchmarked against a PDB-derived testing set and validated against experimental data. Chapter 12 implements a deep learning (DL) platform that teaches drugs to target proteins. This is accomplished by introducing modifications in chemical scaffolds guided by a priori predictions of drug-induced binding-competent conformations of the intended target. In this way, we deal with the fact that target proteins are usually not fixed targets fitting a lock-key paradigm: they structurally adapt to ligands in ways that need to be predicted to empower drug discovery. In contrast with protein folding predictors that do not include disorder propensities as input, the neural network presented solves a “drug-induced folding problem” by incorporating as input sequence-derived signals for structural disorder. The DL system thus predicts conformations in floppy regions of the target protein that rely on associations with a conformation-selecting purposely designed drug to acquire and maintain their structural integrity. This is tantamount to generate within an AI-empowered drug discovery platform the a-priori drug-induced folding ensemble without committing to a specific ligand and then steer the modification of a lead compound to select a specific drug-induced conformation from within the ensemble. Finally, based on the knowledge accrued from the 12 chapters, the Epilogue assesses the plausibility of a futuristic scenario where AI develops its own pattern-based physics.