We start the QUEST,
A.I.-based drug discovery with Quantum Mechanics.
from Small Compounds to Peptide Drugs
‘Deep Computation for Deep Chemistry’. We have pioneered a new drug development platform called 'QUEST', which uses an Artificial Neural Network-based prediction model by adopting the 3D distribution of electrostatic potentials (ESPs) derived from quantum mechanics. QUEST makes it possible to automatically explore optimization of new drug candidates, and even predict chemical properties such as Permeability, Solubility, Stability without wet experiments.
Physics, Chemistry and AI
Physics, chemistry and AI are all necessary for developing new drugs because they provide the knowledge and tools to understand the molecular structures and interactions of drugs, as well as to predict their effects and optimize their design. Physics helps to understand the physical properties of drugs, such as their size, shape, and solubility, which are important for their effectiveness and safety. Chemistry provides the knowledge of chemical reactions and interactions, which are essential for drug synthesis and modification. AI enables the use of advanced computational techniques to analyze large amounts of data and predict the effects of drugs on the body, as well as to optimize their design for specific targets. All of these fields are essential for the development of new and effective drugs.
A human protein hydroxylase that accepts D-residues
Factor inhibiting hypoxia-inducible factor (FIH) is a 2-oxoglutarate-dependent protein hydroxylase that catalyses C3 hydroxylations of protein residues. We report FIH can accept (D)- and (L)-residues for hydroxylation. The substrate selectivity of FIH differs for (D) and (L) epimers, e.g., (D)- but not (L)-allylglycine, and conversely (L)- but not (D)-aspartate, undergo monohydroxylation, in the tested sequence context. The (L)-Leu-containing substrate undergoes FIH-catalysed monohydroxylation, whereas (D)-Leu unexpectedly undergoes dihydroxylation. Crystallographic, mass spectrometric, and DFT studies provide insights into the selectivity of FIH towards (L)- and (D)-residues. The results of this work expand the potential range of known substrates hydroxylated by isolated FIH and imply that it will be possible to generate FIH variants with altered selectivities.
Unraveling innate substrate control in site-selective palladium-catalyzed C–H heterocycle functionalization
The C–H activation of these substrates proceeds via a CMD mechanism, which favors more electron rich positions and therefore displays a pronounced kinetic selectivity for the C3-position. However, C2-selective carbopalladation is also a competitive pathway for chromones so that the overall regiochemical outcome depends on which substrate undergoes activation first. Our studies provide insight into the site-selectivity based on the favorability of two competing CMD and carbopalladation processes of the substrates undergoing coupling. This model can be utilized to predict the regioselectivity of coumarins which are proficient substrates for carbopalladation. Furthermore, our model is able to account for the opposite selectivities observed for enaminone and chromone, and explains how a less reactive coupling partner leads to a switch in selectivity.
Predicting the Electrochemical Properties of Lithium-Ion Battery Electrode Materials with the Quantum Neural Network Algorithm
Discovery of new inorganic solid materials can be accelerated with the aid of a reliable computational tool for predicting the associated electrochemical properties. Hence, we propose a quantitative structure−property relationship model by combining the three-dimensional (3D) quantum mechanical descriptors of materials and the artificial neural network algorithm, which is termed the 3D-QANN model. 3D distribution of electrostatic potentials (ESPs) in the super cell of each inorganic solid material serves as the unique numerical descriptor to derive the 3D-QANN model. The optimized prediction model is then validated in terms of estimating the discharge energy density (D) and the capacity fading (Q) of lithium-ion battery (LIB) cathode materials with the layered structure.
Confinement-Induced Glassy Dynamics in a Model for Chromosome Organization
Recent experiments showing scaling of the intrachromosomal contact probability, P(s)∼s^-1 with the genomic distance s, are interpreted to mean a self-similar fractal-like chromosome organization. The universal value of ϕ∞_c ≈ 0.44 for a sufficiently long polymer (N ≫ 1) allows us to discuss genome dynamics using ϕ as the sole parameter. Our study shows that the on set of glassy dynamics is the reason for the segregated chromosome organization in humans (N ≈ 3 × 109, ϕ≳ϕ∞_c ), whereas chromosomes of budding yeast (N ≈ 108, ϕ < ϕ∞_c ) are equilibrated with no clear signature of such organization.
Emerging β‑Sheet Rich Conformations in Supercompact Huntingtin Exon‑1 Mutant Structures
Here we apply extensive molecular dynamics simulations to study the folding of Htt-exon-1 across five different polyQ-lengths. We find an increase in secondary structure motifs at longer Q-lengths, including β-sheet content that seems to contribute to the formation of increasingly compact structures. More strikingly, these longer Q-lengths adopt supercompact structures as evidenced by a surprisingly small power-law scaling exponent (0.22) between the radius-of-gyration and Q-length that is substantially below expected values for compact globule structures (∼0.33) and unstructured proteins (∼0.50).
Unexpected Swelling of Stiff DNA in a Polydisperse Crowded Environment
We investigate the conformations of DNA-like stiff chains, characterized by contour length (L) and persistence length (lp), in a variety of crowded environments containing monodisperse soft spherical (SS) and spherocylindrical (SC) particles, a mixture of SS and SC, and a milieu mimicking the composition of proteins in the Escherichia coli cytoplasm. The stiff chain, whose size modestly increases in SS crowders up to ϕ ≈ 0.1, is considerably more compact at low volume fractions (ϕ ≤ 0.2) in monodisperse SC particles than in a medium containing SS particles. A 1:1 mixture of SS and SC crowders induces greater chain compaction than the pure SS or SC crowders at the same ϕ, with the effect being highly nonadditive.