Harnessing Shannon entropy-based descriptors in machine learning models to enhance the prediction accuracy of molecular properties.

Accurate prediction of molecular properties is essential in the screening and development of drug molecules and other functional materials. Traditionally, property-specific molecular descriptors are used in machine learning models. This in turn requires the identification and development of target or problem-specific descriptors. Additionally, an increase in the prediction accuracy of the model is not always feasible from the standpoint of targeted descriptor usage. We explored the accuracy and generalizability issues using a framework of Shannon entropies, based on SMILES, SMARTS and/or InChiKey strings of respective molecules. Using various public databases of molecules, we showed that the accuracy of the prediction of machine learning models could be significantly enhanced simply by using Shannon entropy-based descriptors evaluated directly from SMILES. Analogous to partial pressures and total pressure of gases in a mixture, we used atom-wise fractional Shannon entropy in combination with total Shannon entropy from respective tokens of the string representation to model the molecule efficiently. The proposed descriptor was competitive in performance with standard descriptors such as Morgan fingerprints and SHED in regression models. Additionally, we found that either a hybrid descriptor set containing the Shannon entropy-based descriptors or an optimized, ensemble architecture of multilayer perceptrons and graph neural networks using the Shannon entropies was synergistic to improve the prediction accuracy. This simple approach of coupling the Shannon entropy framework to other standard descriptors and/or using it in ensemble models could find applications in boosting the performance of molecular property predictions in chemistry and material science.

Positive and negative chemotaxis of enzyme-coated liposome motors.

The ability of cells or cell components to move in response to chemical signals is critical for the survival of living systems. This motion arises from harnessing free energy from enzymatic catalysis. Artificial model protocells derived from phospholipids and other amphiphiles have been made and their enzymatic-driven motion has been observed. However, control of directionality based on chemical cues (chemotaxis) has been difficult to achieve. Here we show both positive or negative chemotaxis of liposomal protocells. The protocells move autonomously by interacting with concentration gradients of either substrates or products in enzyme catalysis, or Hofmeister salts. We hypothesize that the propulsion mechanism is based on the interplay between enzyme-catalysis-induced positive chemotaxis and solute-phospholipid-based negative chemotaxis. Controlling the extent and direction of chemotaxis holds considerable potential for designing cell mimics and delivery vehicles that can reconfigure their motion in response to environmental conditions.

7 Log Virus Removal in a Simple Functionalized Sand Filter.

Viral contamination of drinking water due to fecal contamination is difficult to detect and treat effectively, leading to frequent outbreaks worldwide. The purpose of this paper is to report on the molecular mechanism for unprecedented high virus removal from a practical sand filter. Sand filters functionalized using a water extract of Moringa oleifera (MO) seeds, functionalized sand (f-sand) filters, achieved a ∼7 log10 virus removal. These tests were conducted with MS2 bacteriophage, a recognized surrogate for pathogenic norovirus and rotavirus. We studied the molecular mechanism of this high removal since it can have important implications for sand filtration, the most common water treatment technology worldwide. Our data reveal that the virus removal activity of f-sand is due to the presence of a chitin-binding protein, M. oleifera chitin-binding protein (MoCBP) on f-sand. Standard column experiments were supported by proteomic analysis and molecular docking simulations. Our simulations show that MoCBP binds preferentially to MS2 capsid proteins demonstrating that specific molecular interactions are responsible for enhanced virus removal. In addition, we simplified the process of making f-sand and evinced how it could be regenerated using saline water. At present, no definitive solution exists for the challenge of treating fecally contaminated drinking and irrigation water for viruses without using technologies that demand high energy or chemical consumption. We propose functionalized sand (f-sand) filters as a highly effective, energy-efficient, and practical technology for virus removal applicable to both developing and developed countries.

Motility of Enzyme-Powered Vesicles.

Autonomous nanovehicles powered by energy derived from chemical catalysis have potential applications as active delivery agents. For in vivo applications, it is necessary that the engine and its fuel, as well as the chassis itself, be biocompatible. Enzyme molecules have been shown to display enhanced motility through substrate turnover and are attractive candidates as engines; phospholipid vesicles are biocompatible and can serve as cargo containers. Herein, we describe the autonomous movement of vesicles with membrane-bound enzymes in the presence of the substrate. We find that the motility of the vesicles increases with increasing enzymatic turnover rate. The enhanced diffusion of these enzyme-powered systems was further substantiated in real time by tracking the motion of the vesicles using optical microscopy. The membrane-bound protocells that move by transducing chemical energy into mechanical motion serve as models for motile living cells and are key to the elucidation of the fundamental mechanisms governing active membrane dynamics and cellular movement.

Nonuniform Crowding Enhances Transport.

The cellular cytoplasm is crowded with macromolecules and other species that occupy up to 40% of the available volume. Previous studies have reported that for high crowder molecule concentrations, colloidal tracer particles have a dampened diffusion due to the higher solution viscosity. However, these studies employed uniform distributions of crowder molecules. We report a scenario, previously unexplored experimentally, of increased tracer transport driven by a nonuniform concentration of crowder macromolecules. In gradients of a polymeric crowder, tracer particles undergo transport several times higher than that of their bulk diffusion rate. The direction of the transport is toward regions of lower crowder concentration. Mechanistically, hard-sphere interactions and the resulting volume exclusion between the tracer and crowder increase the effective diffusion by inducing a convective motion of tracers, which we explain through modeling. Strikingly, soft deformable particles show even greater enhancement in transport in crowder gradients compared to similarly sized hard particles. Overall, this demonstration of enhanced transport in nonuniform distributions of crowders is anticipated to clarify aspects of multicomponent intracellular transport.

Shape-directed rotation of homogeneous micromotors via catalytic self-electrophoresis.

The pursuit of chemically-powered colloidal machines requires individual components that perform different motions within a common environment. Such motions can be tailored by controlling the shape and/or composition of catalytic microparticles; however, the ability to design particle motions remains limited by incomplete understanding of the relevant propulsion mechanism(s). Here, we demonstrate that platinum microparticles move spontaneously in solutions of hydrogen peroxide and that their motions can be rationally designed by controlling particle shape. Nanofabricated particles with n-fold rotational symmetry rotate steadily with speed and direction specified by the type and extent of shape asymmetry. The observed relationships between particle shape and motion provide evidence for a self-electrophoretic propulsion mechanism, whereby anodic oxidation and cathodic reduction occur at different rates at different locations on the particle surface. We develop a mathematical model that explains how particle shape impacts the relevant electrocatalytic reactions and the resulting electrokinetic flows that drive particle motion.

A Theory of Enzyme Chemotaxis: From Experiments to Modeling.

Enzymes show two distinct transport behaviors in the presence of their substrates in solution. First, their diffusivity enhances with an increasing substrate concentration. In addition, enzymes perform directional motion toward regions with a high substrate concentration, termed as chemotaxis. While a variety of enzymes has been shown to undergo chemotaxis, there remains a lack of quantitative understanding of the phenomenon. Here, we derive a general expression for the active movement of an enzyme in a concentration gradient of its substrate. The proposed model takes into account both the substrate-binding and catalytic turnover step, as well as the enhanced diffusion of the enzyme. We have experimentally measured the chemotaxis of a fast and a slow enzyme: urease under catalytic conditions and hexokinase for both full catalysis and for simple noncatalytic substrate binding. There is good agreement between the proposed model and the experiments. The model is general, has no adjustable parameters, and only requires three experimentally defined constants to quantify chemotaxis: enzyme-substrate binding affinity ( Kd), Michaelis-Menten constant ( KM), and level of diffusion enhancement in the associated substrate (α).

Moringa oleifera Seed Protein Adsorption to Silica: Effects of Water Hardness, Fractionation, and Fatty Acid Extraction.

Motivated by the proposed use of cationic protein-modified sand for water filtration in developing nations, this study concerns the adsorption of Moringa oleifera seed proteins to silica surfaces. These proteins were prepared in model waters of varying hardness and underwent different levels of fractionation, including fatty acid extraction and cation exchange chromatography. Adsorption isotherms were measured by ellipsometry, and the zeta potentials of the resulting protein-decorated surfaces were measured by the rotating disk streaming potential method. The results indicate that the presence of fatty acids has little effect on the M. oleifera cationic protein adsorption isotherm. Adsorption from the unfractionated extract was indistinguishable from that of the cationic protein isolates at low concentrations but yielded significantly greater extents of adsorption at high concentrations. Adsorption isotherms for samples prepared in model hard and soft fresh waters were indistinguishable from each other over the measured bulk solution concentration range, but adsorption from hard or soft water was more extensive than adsorption from deionized water at moderate protein concentrations. Streaming potential measurements showed that adsorption reversed the net sign of the zeta potential of silica from negative to positive for all protein fractions and water hardness conditions at protein bulk concentrations as low as 0.03 µg/mL. This suggests that sands can be effectively modified with M. oleifera proteins using small amounts of seed extract under various local water hardness conditions. Finally, ellipsometry indicated that M. oleifera proteins adsorb irreversibly with respect to rinsing in these model fresh waters, suggesting that the modified sand would be stable on repeated use for water filtration. These studies may aid in the design of a simple, effective, and sustainable water purification device for developing nations.

Modulation of Spatiotemporal Particle Patterning in Evaporating Droplets: Applications to Diagnostics and Materials Science.

Spatiotemporal particle patterning in evaporating droplets lacks a common design framework. Here, we demonstrate autonomous control of particle distribution in evaporating droplets through the imposition of a salt-induced self-generated electric field as a generalized patterning strategy. Through modeling, a new dimensionless number, termed "capillary-phoresis" (CP) number, arises, which determines the relative contributions of electrokinetic and convective transport to pattern formation, enabling one to accurately predict the mode of particle assembly by controlling the spontaneous electric field and surface potentials. Modulation of the CP number allows the particles to be focused in a specific region in space or distributed evenly. Moreover, starting with a mixture of two different particle types, their relative placement in the ensuing pattern can be controlled, allowing coassemblies of multiple, distinct particle populations. By this approach, hypermethylated DNA, prevalent in cancerous cells, can be qualitatively distinguished from normal DNA of comparable molecular weights. In other examples, we show uniform dispersion of several particle types (polymeric colloids, multiwalled carbon nanotubes, and molecular dyes) on different substrates (metallic Cu, metal oxide, and flexible polymer), as dictated by the CP number. Depending on the particle, the highly uniform distribution leads to surfaces with a lower sheet resistance, as well as superior dye-printed displays.

Chemotaxis of Molecular Dyes in Polymer Gradients in Solution.

Chemotaxis provides a mechanism for directing the transport of molecules along chemical gradients. Here, we show the chemotactic migration of dye molecules in response to the gradients of several different neutral polymers. The magnitude of chemotactic response depends on the structure of the monomer, polymer molecular weight and concentration, and the nature of the solvent. The mechanism involves cross-diffusion up the polymer gradient, driven by favorable dye-polymer interaction. Modeling allows us to quantitatively evaluate the strength of the interaction and the effect of the various parameters that govern chemotaxis.

Particle Zeta Potentials Remain Finite in Saturated Salt Solutions.

The zeta potential of a particle characterizes its motion in an electric field and is often thought to be negligible at high ionic strength (several moles per liter) due to thinning of the electrical double layer (EDL). Here, we describe zeta potential measurements on polystyrene latex (PSL) particles at monovalent salt concentrations up to saturation (∼5 M NaCl) using electrophoresis in sinusoidal electric fields and high-speed video microscopy. Our measurements reveal that the zeta potential remains finite at even the highest concentrations. Moreover, we find that the zeta potentials of sulfated PSL particles continue to obey the classical Gouy-Chapman model up to saturation despite significant violations in the model's underlying assumptions. By contrast, amidine-functionalized PSL particles exhibit qualitatively different behaviors such as zero zeta potentials at high concentrations of NaCl and KCl and even charge inversion in KBr solutions. The experimental results are reproduced and explained by Monte Carlo simulations of a simple lattice model of the EDL that accounts for effects due to ion size and ion-ion correlations. At high salt conditions, the model suggests that quantitative changes in the magnitude of surface charge can result in qualitative changes in the zeta potential-most notably, charge inversion of highly charged surfaces. These findings have important implications for electrokinetic phenomena such as diffusiophoresis within salty environments such as oceans, geological reservoirs, and living organisms.

Origins of concentration gradients for diffusiophoresis.

Fluid transport that is driven by gradients of pressure, gravity, or electro-magnetic potential is well-known and studied in many fields. A subtler type of transport, called diffusiophoresis, occurs in a gradient of chemical concentration, either electrolyte or non-electrolyte. Diffusiophoresis works by driving a slip velocity at the fluid-solid interface. Although the mechanism is well-known, the diffusiophoresis mechanism is often considered to be an esoteric laboratory phenomenon. However, in this article we show that concentration gradients can develop in a surprisingly wide variety of physical phenomena - imposed gradients, asymmetric reactions, dissolution, crystallization, evaporation, mixing, sedimentation, and others - so that diffusiophoresis is in fact a very common transport mechanism, in both natural and artificial systems. We anticipate that in georeservoir extractions, physiological systems, drying operations, laboratory and industrial separations, crystallization operations, membrane processes, and many other situations, diffusiophoresis is already occurring - often without being recognized - and that opportunities exist for designing this transport to great advantage.

Self-Generated Electrokinetic Fluid Flows during Pseudomorphic Mineral Replacement Reactions.

Pseudomorphic mineral replacement reactions involve one mineral phase replacing another, while preserving the original mineral's size and texture. Macroscopically, these transformations are driven by system-wide equilibration through dissolution and precipitation reactions. It is unclear, however, how replacement occurs on the molecular scale and what role dissolved ion transport plays. Here, we develop a new quantitative framework to explain the pseudomorphic replacement of KBr crystal in a saturated KCl solution through a combination of microscopic, spectroscopic, and modeling techniques. Our observations reveal that pseudomorphic mineral replacement (pMRR) is transport-controlled for this system and that convective fluid flows, caused by diffusioosmosis, play a key role in the ion transport process across the reaction-induced pores in the product phase. Our findings have important implications for understanding mineral transformations in natural environments and suggest that replacement could be exploited in commercial and laboratory applications.

Boundaries can steer active Janus spheres.

The advent of autonomous self-propulsion has instigated research towards making colloidal machines that can deliver mechanical work in the form of transport, and other functions such as sensing and cleaning. While much progress has been made in the last 10 years on various mechanisms to generate self-propulsion, the ability to steer self-propelled colloidal devices has so far been much more limited. A critical barrier in increasing the impact of such motors is in directing their motion against the Brownian rotation, which randomizes particle orientations. In this context, here we report directed motion of a specific class of catalytic motors when moving in close proximity to solid surfaces. This is achieved through active quenching of their Brownian rotation by constraining it in a rotational well, caused not by equilibrium, but by hydrodynamic effects. We demonstrate how combining these geometric constraints can be utilized to steer these active colloids along arbitrary trajectories.

Enhanced transport into and out of dead-end pores.

Dead-end micro- and nanoscale channels are ubiquitous in nature and are found in geological and biological systems subject to frequent disruptions. Achieving fluid flows in them is not possible through conventional pressure-driven mechanisms. Here we show that chemically driven convective flows leading to transport in and out of dead-end pores can occur by the phenomenon of "transient diffusioosmosis". The advective velocity depends on the presence of an in situ-generated transient ion gradient and the intrinsic charge on the pore wall. The flows can reach speeds of 50 µm/s and cause extraction of otherwise-trapped materials. Our results illustrate that chemical energy, in the form of a transient salt gradient, can be transduced into mechanical motion with the pore wall acting as the pump. As discussed, the phenomena may underlie observed transport in many geological and biological systems involving tight or dead-end micro- and nanochannels.

Multi-ion diffusiophoresis.

The movement of charged particles occurs in a salt concentration gradient by the mechanism of diffusiophoresis. Current analytical models for diffusiophoresis have been developed for a gradient generated by a symmetric Z:Z electrolyte at steady state. Recently, our lab has reported diffusioosmotic flows due to dissolving calcium carbonate (CaCO3) which generates three ions (Ca(2+), HCO3(-), and OH(-)), and this fluid motion cannot be described by current analytical models. In this communication, we derive an expression for the diffusioosmotic flow in a gradient involving multiple ions of different valences, assuming infinitesimally thin double layers. We also solve numerically the time-dependent concentration profiles for the three-ion case and find that the concentration profile of HCO3(-) is non-monotonic in solution. Finally, we examine quantitatively the assumption of electroneutrality in solution, finding that electroneutrality is a good approximation even for the multi-ion case, indicating that our electric field derived from the ion migration equation is quite accurate.

Localized electroosmosis (LEO) induced by spherical colloidal motors.

Experimental data show that the speed of colloidal (catalytic) motors decreases as the size of the motor particles increases. However, previous electrokinetic models have shown that the colloidal motor speed for spheres is independent of size, at least for the case of infinitesimally thin double layers and reaction-limited catalysis. Although a size dependence of motor speed has been calculated for diffusion-limited catalysis, most motor experiments are done in the reaction-limited regime. This apparent contradiction led us to examine how motor speed (U) changes with distance (δ) from a wall, starting from the usual electrokinetic equations. A key finding is that interactions between a colloidal motor and a nearby wall produce a localized electroosmotic (LEO) flow field that can significantly alter the motor speed near the wall. Because large motor particles typically settle closer to the wall than small motors, LEO thus provides at least one explanation of the size dependence of motor speed. Furthermore, LEO provides a new method of creating flow fields in capillaries and microchannels.

Particle deposition on microporous membranes can be enhanced or reduced by salt gradients.

Colloidal particle deposition on membranes is a continuing scientific and technological challenge. In this paper we examine the role of a previously unexplored phenomenon-diffusiophoretic particle transport toward a membrane-in relation to fouling. Diffusiophoresis is an electrokinetic transport mechanism that arises in salt gradients, especially when the ions have different diffusion coefficients. Through experiments conducted with salt diffusing across microdialysis membranes, with no advection, we show experimentally that diffusiophoresis induces colloidal deposition on the surface of microporous surfaces. We used transient salt (NaCl, KCl, LiCl) gradients and fundamental electrokinetic modeling to assess the role of diffusiophoresis in colloidal fouling. Based on (i) difference in diffusion coefficients of ions, (ii) zeta potential on the particles, and (iii) ionic gradient applied across the walls of the membrane, colloidal fouling could be both quantitatively and qualitatively predicted. Our understanding enabled us to stop particle deposition by adding calcium carbonate outside the membrane, which generates a stronger electric field in a direction opposite to that created by salt diffusing from the membrane. We propose that accounting for this diffusiophoretic mode of particle deposition is important in understanding membrane fouling.

Polloidal chains from self-assembly of flattened particles.

Chains of micrometer-size colloidal particles have been self-assembled that are flexible, mechanically stable, and observable in optical microscopy. The chains sometimes have more than 30 particles, and we call them "polloidal chains". A key aspect of the work is the careful modeling of the interparticle forces between partially flattened polystyrene spheres. This modeling helped us to identify a narrow window of system conditions that produce interparticle physical bonds with a bond energy greater than 15kT, as well as a gap of fluid between particles that enables freely rotating bonds and flexible chains. The formation of the chains is well-modeled using linear condensation growth from classical polymer theory, suggesting that the chains might be used experimentally as large-scale, relatively slow moving models for polymer chains.