Trustworthy in silico cell labeling via ensemble-based image translation

Artificial intelligence (AI) image translation has been a valuable tool for processing image data in biological and medical research. To apply such a tool in mission-critical applications, including drug screening, toxicity study, and clinical diagnostics, it is essential to ensure that the AI prediction is trustworthy. Here, we demonstrate that an ensemble learning method can quantify the uncertainty of AI image translation. We tested the uncertainty evaluation using experimentally acquired images of mesenchymal stromal cells. We find that the ensemble method reports a prediction standard deviation that correlates with the prediction error, estimating the prediction uncertainty. We show that this uncertainty is in agreement with the prediction error and Pearson correlation coefficient. We further show that the ensemble method can detect out-of-distribution input images by reporting increased uncertainty. Altogether, these results suggest that the ensemble-estimated uncertainty can be a useful indicator for identifying erroneous AI image translations.

Prostate lineage-specific metabolism governs luminal differentiation and response to antiandrogen treatment

Lineage transitions are a central feature of prostate development, tumourigenesis and treatment resistance. While epigenetic changes are well known to drive prostate lineage transitions, it remains unclear how upstream metabolic signalling contributes to the regulation of prostate epithelial identity. To fill this gap, we developed an approach to perform metabolomics on primary prostate epithelial cells. Using this approach, we discovered that the basal and luminal cells of the prostate exhibit distinct metabolomes and nutrient utilization patterns. Furthermore, basal-to-luminal differentiation is accompanied by increased pyruvate oxidation. We establish the mitochondrial pyruvate carrier and subsequent lactate accumulation as regulators of prostate luminal identity. Inhibition of the mitochondrial pyruvate carrier or supplementation with exogenous lactate results in large-scale chromatin remodelling, influencing both lineage-specific transcription factors and response to antiandrogen treatment. These results establish reciprocal regulation of metabolism and prostate epithelial lineage identity.

Microwell-based flow culture increases viability and restores drug response in prostate cancer spheroids

3D cancer spheroids represent a highly promising model for study of cancer progression and therapeutic development. Wide-scale adoption of cancer spheroids, however, remains a challenge due to the lack of control over hypoxic gradients that may cloud the assessment of cell morphology and drug response. Here, we present a Microwell Flow Device (MFD) that generates in-well laminar flow around 3D tissues via repetitive tissue sedimentation. Using a prostate cancer cell line, we demonstrate the spheroids in the MFD exhibit improved cell growth, reduced necrotic core formation, enhanced structural integrity, and downregulated expression of cell stress genes. The flow-cultured spheroids also exhibit an improved sensitivity to chemotherapy with greater transcriptional response. These results demonstrate how fluidic stimuli reveal the cellular phenotype previously masked by severe necrosis. Our platform advances 3D cellular models and enables study into hypoxia modulation, cancer metabolism, and drug screening within pathophysiological conditions.

Investigating heterogeneities of live mesenchymal stromal cells using AI-based label-free imaging

Mesenchymal stromal cells (MSCs) are multipotent cells that have great potential for regenerative medicine, tissue repair, and immunotherapy. Unfortunately, the outcomes of MSC-based research and therapies can be highly inconsistent and difficult to reproduce, largely due to the inherently significant heterogeneity in MSCs, which has not been well investigated. To quantify cell heterogeneity, a standard approach is to measure marker expression on the protein level via immunochemistry assays. Performing such measurements non-invasively and at scale has remained challenging as conventional methods such as flow cytometry and immunofluorescence microscopy typically require cell fixation and laborious sample preparation. Here, we developed an artificial intelligence (AI)-based method that converts transmitted light microscopy images of MSCs into quantitative measurements of protein expression levels. By training a U-Net+ conditional generative adversarial network (cGAN) model that accurately (mean [Formula: see text] = 0.77) predicts expression of 8 MSC-specific markers, we showed that expression of surface markers provides a heterogeneity characterization that is complementary to conventional cell-level morphological analyses. Using this label-free imaging method, we also observed a multi-marker temporal-spatial fluctuation of protein distributions in live MSCs. These demonstrations suggest that our AI-based microscopy can be utilized to perform quantitative, non-invasive, single-cell, and multi-marker characterizations of heterogeneous live MSC culture. Our method provides a foundational step toward the instant integrative assessment of MSC properties, which is critical for high-throughput screening and quality control in cellular therapies.

Computing the self-consistent field in Kohn-Sham density functional theory

A new framework is presented for evaluating the performance of self-consistent field methods in Kohn-Sham density functional theory (DFT). The aims of this work are two-fold. First, we explore the properties of Kohn-Sham DFT as it pertains to the convergence of self-consistent field iterations. Sources of inefficiencies and instabilities are identified, and methods to mitigate these difficulties are discussed. Second, we introduce a framework to assess the relative utility of algorithms in the present context, comprising a representative benchmark suite of over fifty Kohn-Sham simulation inputs, the scf-x _n suite. This provides a new tool to develop, evaluate and compare new algorithms in a fair, well-defined and transparent manner.

Prediction of GABARAP interaction with the GABA type A receptor

We have performed docking simulations on GABARAP interacting with the GABA type A receptor using SwarmDock. We have also used a novel method to study hydration sites on the surface of these two proteins; this method identifies regions around proteins where desolvation is relatively easy, and these are possible locations where proteins can bind each other. There is a high degree of consistency between the predictions of these two methods. Moreover, we have also identified binding sites on GABARAP for other proteins, and listed possible binding sites for as yet unknown proteins on both GABARAP and the GABA type A receptor intracellular domain.

Predicting solvatochromic shifts and colours of a solvated organic dye: The example of nile red

The solvatochromic shift, as well as the change in colour of the simple organic dye nile red, is studied in two polar and two non-polar solvents in the context of large-scale time-dependent density-functional theory (TDDFT) calculations treating large parts of the solvent environment from first principles. We show that an explicit solvent representation is vital to resolve absorption peak shifts between nile red in n-hexane and toluene, as well as acetone and ethanol. The origin of the failure of implicit solvent models for these solvents is identified as being due to the strong solute-solvent interactions in form of π-stacking and hydrogen bonding in the case of toluene and ethanol. We furthermore demonstrate that the failures of the computationally inexpensive Perdew-Burke-Ernzerhof (PBE) functional in describing some features of the excited state potential energy surface of the S₁ state of nile red can be corrected for in a straightforward fashion, relying only on a small number of calculations making use of more sophisticated range-separated hybrid functionals. The resulting solvatochromic shifts and predicted colours are in excellent agreement with experiment, showing the computational approach outlined in this work to yield very robust predictions of optical properties of dyes in solution.

Simulation of electron energy loss spectra of nanomaterials with linear-scaling density functional theory

Experimental techniques for electron energy loss spectroscopy (EELS) combine high energy resolution with high spatial resolution. They are therefore powerful tools for investigating the local electronic structure of complex systems such as nanostructures, interfaces and even individual defects. Interpretation of experimental electron energy loss spectra is often challenging and can require theoretical modelling of candidate structures, which themselves may be large and complex, beyond the capabilities of traditional cubic-scaling density functional theory. In this work, we present functionality to compute electron energy loss spectra within the onetep linear-scaling density functional theory code. We first demonstrate that simulated spectra agree with those computed using conventional plane wave pseudopotential methods to a high degree of precision. The ability of onetep to tackle large problems is then exploited to investigate convergence of spectra with respect to supercell size. Finally, we apply the novel functionality to a study of the electron energy loss spectra of defects on the (1 0 1) surface of an anatase slab and determine concentrations of defects which might be experimentally detectable.

Linear-scaling time-dependent density-functional theory beyond the Tamm-Dancoff approximation: Obtaining efficiency and accuracy with in situ optimised local orbitals

We present a solution of the full time-dependent density-functional theory (TDDFT) eigenvalue equation in the linear response formalism exhibiting a linear-scaling computational complexity with system size, without relying on the simplifying Tamm-Dancoff approximation (TDA). The implementation relies on representing the occupied and unoccupied subspaces with two different sets of in situ optimised localised functions, yielding a very compact and efficient representation of the transition density matrix of the excitation with the accuracy associated with a systematic basis set. The TDDFT eigenvalue equation is solved using a preconditioned conjugate gradient algorithm that is very memory-efficient. The algorithm is validated on a small test molecule and a good agreement with results obtained from standard quantum chemistry packages is found, with the preconditioner yielding a significant improvement in convergence rates. The method developed in this work is then used to reproduce experimental results of the absorption spectrum of bacteriochlorophyll in an organic solvent, where it is demonstrated that the TDA fails to reproduce the main features of the low energy spectrum, while the full TDDFT equation yields results in good qualitative agreement with experimental data. Furthermore, the need for explicitly including parts of the solvent into the TDDFT calculations is highlighted, making the treatment of large system sizes necessary that are well within reach of the capabilities of the algorithm introduced here. Finally, the linear-scaling properties of the algorithm are demonstrated by computing the lowest excitation energy of bacteriochlorophyll in solution. The largest systems considered in this work are of the same order of magnitude as a variety of widely studied pigment-protein complexes, opening up the possibility of studying their properties without having to resort to any semiclassical approximations to parts of the protein environment.

Linear-scaling density-functional simulations of charged point defects in Al2O3 using hierarchical sparse matrix algebra

We present calculations of formation energies of defects in an ionic solid (Al(2)O(3)) extrapolated to the dilute limit, corresponding to a simulation cell of infinite size. The large-scale calculations required for this extrapolation are enabled by developments in the approach to parallel sparse matrix algebra operations, which are central to linear-scaling density-functional theory calculations. The computational cost of manipulating sparse matrices, whose sizes are determined by the large number of basis functions present, is greatly improved with this new approach. We present details of the sparse algebra scheme implemented in the ONETEP code using hierarchical sparsity patterns, and demonstrate its use in calculations on a wide range of systems, involving thousands of atoms on hundreds to thousands of parallel processes.

Surface diffusion: the low activation energy path for nanotube growth

We present the temperature dependence of the growth rate of carbon nanofibers by plasma-enhanced chemical vapor deposition with Ni, Co, and Fe catalysts. We extrapolate a common low activation energy of 0.23-0.4 eV, much lower than for thermal deposition. The carbon diffusion on the catalyst surface and the stability of the precursor molecules, C2H2 or CH4, are investigated by ab initio plane wave density functional calculations. We find a low activation energy of 0.4 eV for carbon surface diffusion on Ni and Co (111) planes, much lower than for bulk diffusion. The energy barrier for C2H2 and CH4 dissociation is at least 1.3 eV and 0.9 eV, respectively, on Ni(111) planes or step edges. Hence, the rate-limiting step for plasma-enhanced growth is carbon diffusion on the catalyst surface, while an extra barrier is present for thermal growth due to gas decomposition.

"Learn on the fly": a hybrid classical and quantum-mechanical molecular dynamics simulation

We describe and test a novel molecular dynamics method which combines quantum-mechanical embedding and classical force model optimization into a unified scheme free of the boundary region, and the transferability problems which these techniques, taken separately, involve. The scheme is based on the idea of augmenting a unique, simple parametrized force model by incorporating in it, at run time, the quantum-mechanical information necessary to ensure accurate trajectories. The scheme is tested on a number of silicon systems composed of up to approximately 200 000 atoms.

Thermodynamics of catalytic formation of dimethyl ether from methanol in acidic zeolites

We present a theoretical study of the formation of the first intermediate, dimethyl ether, in the methanol to gasoline conversion within the framework of an ab initio molecular dynamics approach. The study is performed under conditions that closely resemble the reaction conditions in the zeolite catalyst including the full topology of the framework. The use of the method of thermodynamic integration allows us to extract the free-energy profile along the reaction coordinate. We find that the entropic contribution qualitatively alters the free-energy profile relative to the total energy profile. Different transition states are found from the internal and free energy profiles. The entropy contribution varies significantly along the reaction coordinate and is responsible for stabilizing the products and for lowering the energy barrier. The hugely inhomogeneous variation of the entropy can be understood in terms of elementary processes that take place during the chemical reaction. Our simulations provide new insights into the complex nature of this chemical reaction.

Can atomic force microscopy achieve atomic resolution in contact mode?

Atomic force microscopy operating in the contact mode is studied using total-energy pseudopotential calculations. It is shown that, in the case of a diamond tip and a diamond surface, it is possible for a tip terminated by a single atom to sustain forces in excess of 30 nN. It is also shown that imaging at atomic resolution may be limited by blunting of the tip during lateral scanning.

Structural properties of lanthanide and actinide compounds within the plane wave pseudopotential approach

We show that plane wave ultrasoft pseudopotential methods readily extend to the calculation of the structural properties of lanthanide and actinide containing compounds. This is demonstrated through a series of calculations performed on UO, UO2, UO3, U3O8, UC2, alpha-CeC2, CeB6, CeSe, CeO2, NdB6, TmOI, LaBi, LaTiO3, YbO, and elemental Lu.

Theoretical strength and cleavage of diamond

The theoretical strength of diamond has been calculated for the <100>, <110>, and <111> directions using a first principles approach and is found to be strongly dependent on crystallographic direction. This elastic anisotropy, found at large strains, and particularly the pronounced minimum in cohesion in the <111> direction, is believed to be the reason for the remarkable dominance of the 111 cleavage plane when diamond is fractured. The extra energy required to cleave a crystal on planes other than 111 is discussed with reference to simple surface energy calculations and also the introduction of bond-bending terms.

First principles investigation of singly reduced cytochrome P450

1. The application of novel ab initio quantum mechanical methods to the states in the catalytic cycle of cytochrome P450 following the first reduction step is described. 2. A good correlation was found between the calculated energy of reduction and the experimentally determined redox potential for a range of substrate- and substrate analogue-bound systems. 3. On reduction of the haem system, the ground state of Fe remains Fe3+. On binding of a CO molecule, Fe adopts a low-spin Fe2+ state, in agreement with experiment. However, on binding of an O2 molecule, calculations indicate that the system adopts a ferric superoxide ground state, in which the Fe is in a low-spin Fe3+ state.

Evidence for stabilization of the low-spin state of cytochrome P450 due to shortening of the proximal heme bond

The cytochrome P450 superfamily of enzymes is ubiquitous, being responsible for the metabolism of a wide range of endogenous and xenobiotic compounds. However, the detailed mechanism of the catalytic cycle of these enzymes is still not fully understood. We describe results, obtained from first principles molecular simulations, which indicate that the low-spin state of the Fe3+ ion, present in the heme moiety at the active site of a cytochrome P450 enzyme, may be stabilized by shortening of the proximal bond of the heme. Calculations indicate that a bond length of less than approximately 2.05 A between the heme Fe3+ ion and the cysteine S, which forms the proximal ligand, would result in the stabilization of the low-spin state of the Fe3+, inhibiting the progress of the P450 catalytic cycle. Our investigation uses novel first principles modeling techniques which treat the entire system quantum-mechanically.

An ab initio approach to the understanding of cytochrome P450-ligand interactions

1. We describe the application of novel ab initio quantum mechanical methods to the study of ligand interactions with cytochrome P450cam (CYP101). 2. We find that our techniques accurately describe the transition from a low-spin state to a high-spin state of the haem Fe3+ on binding of a substrate. Furthermore, our methods correctly predict that a large fraction of low-spin character is retained on binding of an inhibitor. 3. We demonstrate the use of 'computational experiments' to elucidate key features of the mechanism of interaction. This leads us to identify a new mechanism for the suppression of the low- to high-spin transition on binding of an inhibitor, namely the shortening of the bond between the Fe atom and the coordinated S atom of the cysteine axial ligand.

Ab initio molecular modeling in the study of drug metabolism

We discuss the use of ab initio quantum mechanical methods in drug metabolism studies. These methods require only the positions and atomic numbers of the atoms to be specified and offer greater transferability than conventional molecular modeling techniques. This fact, coupled with the accuracy of our approach, permits 'computational experiments' to be performed, allowing details of reaction mechanisms to be understood. We review the application of these methods to the cytochrome P450 superfamily of enzymes. There is much interest in understanding the mechanisms of these enzymes due to their participation in a wide range of metabolic processes including drug activation/deactivation. We find that our methods accurately reproduce the low- to high-spin transition of the haem Fe on binding of a substrate. Furthermore, we identify a new mechanism for the suppression of this spin transition, namely the shortening of the bond between the Fe atom and the coordinated S atom of the cysteine axial ligand. These results indicate that ab initio molecular modeling may be usefully applied in the study of drug metabolism and that further study of intermediate states in the P450 reaction cycle would be beneficial, particularly those which are not accessible using conventional experimental approaches.

Nicotine alleviation of nicotine abstinence syndrome is naloxone-reversible

In a recently introduced rodent model of nicotine abstinence syndrome, the observed signs closely resembled those typical of rat opiate abstinence syndrome. Signs were precipitated by naloxone and potently reversed by morphine as well as nicotine itself, suggesting that nicotine might relieve nicotine abstinence syndrome through releasing endogenous opioids. To test this hypothesis, rats were continuously infused subcutaneously (SC) for 7 days with 9 mg/kg per day nicotine tartrate. Each rat was observed for abstinence signs at 18 and 21 h after termination of infusion. Three minutes before the 21-h test, all rats received 0.35 mg/kg nicotine tartrate, SC; 5 min before the nicotine injection, subjects received 9 or 4.5 mg/kg naloxone or saline alone, SC. Abstinence reversal scores were calculated as signs at 21 h as a percentage of signs at 18 h. Naloxone prevented nicotine alleviation of nicotine abstinence in a dose-related manner. However, naloxone in the absence of a nicotine injection had no effect on abstinence severity in either highly dependent or moderately dependent rats (infused with 9 or 5 mg/kg per day nicotine tartrate, respectively). These results support the hypothesis that endogenous opioids play a role in nicotine dependence and abstinence.

The renal transplant perfusion index: reduction in the error and variability

For 15 years, perfusion indices derived from scintigraphic studies have proved useful in the serial evaluation of renal transplants and they have recently been confirmed as being more sensitive than Doppler ultrasound as an indicator of vascular rejection and cyclosporin toxicity. In the calculation of these indices, correction for administered activity is often accomplished using activity measurements made over a convenient artery which, therefore, has a critical influence on the value of the index obtained. In this communication, a theoretical assessment is made of the error and variability introduced into the calculation of the perfusion index, because of inadequate spatial sampling of activity in these narrow arteries and the consequential inconsistencies in the measurement of the arterial tracer activity. Using numerical simulation, it is shown that the errors in repeat studies on the same patient may be as high as 39% and between patients as high as 53%. These figures can be reduced to below 18% and 21%, respectively, by constructing a region of interest (ROI) to extend over as much of the arterial width as possible rather than relying only on the maximum pixel count. Further reduction to below 12% and 10% is possible by utilising a 128 x 128 acquisition matrix instead of 64 x 64 and drawing the ROI over the aorta instead of the iliac artery.