Optimized Bags of Artificial Neural Networks Can Predict the Viability of Organisms Exposed to Nanoparticles.

Prediction of organismal viability upon exposure to a nanoparticle in varying environments─as fully specified at the molecular scale─has emerged as a useful figure of merit in the design of engineered nanoparticles. We build on our earlier finding that a bag of artificial neural networks (ANNs) can provide such a prediction when such machines are trained with a relatively small data set (with ca. 200 examples). Therein, viabilities were predicted by consensus using the weighted means of the predictions from the bags. Here, we confirm the accuracy and precision of the prediction of nanoparticle viabilities using an optimized bag of ANNs over sets of data examples that had not previously been used in the training and validation process. We also introduce the viability strip, rather than a single value, as the prediction and construct it from the viability probability distribution of an ensemble of ANNs compatible with the data set. Specifically, the ensemble consists of the ANNs arising from subsets of the data set corresponding to different splittings between training and validation, and the different bags (k-folds). A k-1k machine uses a single partition (or bag) of k - 1 ANNs each trained on 1/k of the data to obtain a consensus prediction, and a k-bag machine quorum samples the k possible k-1k machines available for a given partition. We find that with increasing k in the k-bag or k-1k machines, the viability strips become more normally distributed and their predictions become more precise. Benchmark comparisons between ensembles of 4-bag machines and 34 fraction machines suggest that the 34 fraction machine has similar accuracy while overcoming some of the challenges arising from divergent ANNs in the 4-bag machines.

Development of a universal method for vibrational analysis of the terminal alkyne C≡C stretch.

The terminal alkyne C≡C stretch has a large Raman scattering cross section in the "silent" region for biomolecules. This has led to many Raman tag and probe studies using this moiety to study biomolecular systems. A computational investigation of these systems is vital to aid in the interpretation of these results. In this work, we develop a method for computing terminal alkyne vibrational frequencies and isotropic transition polarizabilities that can easily and accurately be applied to any terminal alkyne molecule. We apply the discrete variable representation method to a localized version of the C≡C stretch normal mode. The errors of (1) vibrational localization to the terminal alkyne moiety, (2) anharmonic normal mode isolation, and (3) discretization of the Born-Oppenheimer potential energy surface are quantified and found to be generally small and cancel each other. This results in a method with low error compared to other anharmonic vibrational methods like second-order vibrational perturbation theory and to experiments. Several density functionals are tested using the method, and TPSS-D3, an inexpensive nonempirical density functional with dispersion corrections, is found to perform surprisingly well. Diffuse basis functions are found to be important for the accuracy of computed frequencies. Finally, the computation of vibrational properties like isotropic transition polarizabilities and the universality of the localized normal mode for terminal alkynes are demonstrated.

Engineered Chimeras Unveil Swappable Modular Features of Fatty Acid and Polyketide Synthase Acyl Carrier Proteins.

The strategic redesign of microbial biosynthetic pathways is a compelling route to access molecules of diverse structure and function in a potentially environmentally sustainable fashion. The promise of this approach hinges on an improved understanding of acyl carrier proteins (ACPs), which serve as central hubs in biosynthetic pathways. These small, flexible proteins mediate the transport of molecular building blocks and intermediates to enzymatic partners that extend and tailor the growing natural products. Past combinatorial biosynthesis efforts have failed due to incompatible ACP-enzyme pairings. Herein, we report the design of chimeric ACPs with features of the actinorhodin polyketide synthase ACP (ACT) and of the Escherichia coli fatty acid synthase (FAS) ACP (AcpP). We evaluate the ability of the chimeric ACPs to interact with the E. coli FAS ketosynthase FabF, which represents an interaction essential to building the carbon backbone of the synthase molecular output. Given that AcpP interacts with FabF but ACT does not, we sought to exchange modular features of ACT with AcpP to confer functionality with FabF. The interactions of chimeric ACPs with FabF were interrogated using sedimentation velocity experiments, surface plasmon resonance analyses, mechanism-based cross-linking assays, and molecular dynamics simulations. Results suggest that the residues guiding AcpP-FabF compatibility and ACT-FabF incompatibility may reside in the loop I, α-helix II region. These findings can inform the development of strategic secondary element swaps that expand the enzyme compatibility of ACPs across systems and therefore represent a critical step toward the strategic engineering of "un-natural" natural products.

The Primarily Undergraduate Nanomaterials Cooperative: A New Model for Supporting Collaborative Research at Small Institutions on a National Scale.

The Primarily Undergraduate Nanomaterials Cooperative (PUNC) is an organization for research-active faculty studying nanomaterials at Primarily Undergraduate Institutions (PUIs), where undergraduate teaching and research go hand-in-hand. In this perspective, we outline the differences in maintaining an active research group at a PUI compared to an R1 institution. We also discuss the work of PUNC, which focuses on community building, instrument sharing, and facilitating new collaborations. Currently consisting of 37 members from across the United States, PUNC has created an online community consisting of its Web site (nanocooperative.org), a weekly online summer group meeting program for faculty and students, and a Discord server for informal conversations. Additionally, in-person symposia at ACS conferences and PUNC-specific conferences are planned for the future. It is our hope that in the years to come PUNC will be seen as a model organization for community building and research support at primarily undergraduate institutions.

Optimizing bags of artificial neural networks for the prediction of viability from sparse data.

The prediction of organismal viability in response to exposure to different nanoparticles and conditions characterized at the molecular scale is challenging because several orders of magnitude must be bridged. A so-called bag of artificial neural networks has recently been shown to provide such a connection when trained through the use of relatively small databases. In the present work, we found that individual neural networks do not always converge under training. The use of bags and weighted means for consensus predictions provides a mechanism for effectively pruning the effects from the diverging networks without requiring prior conditioning. The optimized structure of these machines was also found to be effective in predicting the relative viability of nanoparticles that had not been used in developing the major findings of this work.

Surface Coating Structure and Its Interaction with Cytochrome c in EG6-Coated Nanoparticles Varies with Surface Curvature.

The composition, orientation, and conformation of proteins in biomolecular coronas acquired by nanoparticles in biological media contribute to how they are identified by a cell. While numerous studies have investigated protein composition in biomolecular coronas, relatively little detail is known about how the nanoparticle surface influences the orientation and conformation of the proteins associated with them. We previously showed that the peripheral membrane protein cytochrome c adopts preferred poses relative to negatively charged 3-mercaptopropionic acid (MPA)-gold nanoparticles (AuNPs). Here, we employ molecular dynamics simulations and complementary experiments to establish that cytochrome c also assumes preferred poses upon association with nanoparticles functionalized with an uncharged ligand, specifically ω-(1-mercaptounde-11-cyl)hexa(ethylene glycol) (EG6). We find that the display of the EG6 ligands is sensitive to the curvature of the surface-and, consequently, the effective diameter of the nearly spherical nanoparticle core-which in turn affects the preferred poses of cytochrome c.

Modeling Carbon Dioxide Vibrational Frequencies in Ionic Liquids: IV. Temperature Dependence.

In previous papers in the series, the vibrational spectroscopy of CO2 in ionic liquids (ILs) was investigated at ambient conditions. Here, we extend these studies to understand the temperature dependence of the structure, dynamics, and thermodynamics of CO2 in the 1-butyl-3-methylimidazolium hexafluorophosphate, [bmim][PF6], IL. Using spectroscopic mapping techniques, the infrared absorption spectrum of the CO2 asymmetric stretch mode is simulated at a number of temperatures, and the results are found to be consistent with similar experimental studies. Structural correlation functions are used to reveal the thermodynamics of complete CO2 solvent cage breakdown. The enthalpy and entropy of activation for solvent cage reorganization are found to be 6.9 and 7.6 (kcal/mol)/K, respectively, and these values are similar to the those for spectral, orientational, and translational diffusion. Caging times for CO2 are calculated, and it is shown that the short time dynamics of CO2 are unaffected by temperature, even though the long-time dynamics are highly sensitive to temperature.

Counter Cations Affect Transport in Aqueous Hydroxide Solutions with Ion Specificity.

The anomalously high mobility of hydroxide and hydronium ions in aqueous solutions is related to proton transfer and structural diffusion. The role of counterions in these solutions, however, is often considered to be negligible. Herein, we explore the impact of alkali metal counter cations on hydroxide solvation and mobility. Impedance measurements demonstrate that hydroxide mobility is attenuated by lithium relative to sodium and potassium. These results are explained by ab initio molecular dynamics simulations and experimental vibrational hydration shell spectroscopy, which reveal substantially stronger ion pairing between OH- and Li+ than with other cations. Hydration shell spectra and theoretical vibrational frequency calculations together imply that lithium and sodium cations have different effects on the delocalization of water protons donating a hydrogen bond to hydroxide. Specifically, lithium leads to enhanced proton delocalization compared with sodium. However, proton delocalization and the overall diffusion process are not necessarily correlated.

Modeling Carbon Dioxide Vibrational Frequencies in Ionic Liquids: III. Dynamics and Spectroscopy.

In recent years, interest in carbon capture and sequestration has led to numerous investigations of the ability of ionic liquids to act as recyclable CO2-sorbent materials. Herein, we investigate the structure and dynamics of a model physisorbing ionic liquid, 1-butyl-3-methylimidazolium hexafluorophosphate ([C4C1Im][PF6]), from the perspective of CO2 using two-dimensional (2D) IR spectroscopy and molecular dynamics simulations. A direct comparison of experimentally measured and calculated 2D IR line shapes confirms the validity of the simulations and spectroscopic calculations. Taken together, the simulations and experiments reveal new insights into the interactions of a CO2 solute with the surrounding ionic liquid and how these interactions manifest in the 2D IR spectra. In particular, higher CO2 asymmetric stretch vibrational frequencies are associated with softer, less populated solvent cages and lower frequencies are associated with tighter, more highly populated solvent cages. The CO2 interacts most strongly with the anions, and these interactions persist for more than 1 ns. The second strongest interactions are with the imidazolium cation ring that last 100 ps, and the weakest interactions are with the cation butyl tail that persist for 10 ps. The principal contributors to spectral diffusion of the CO2 asymmetric stretch vibrational frequency due to the dynamical evolution of the solvent are through Lennard-Jones interactions at short times and electrostatics at long times.

Enthalpic Driving Force for the Selective Absorption of CO2 by an Ionic Liquid.

Molecular dynamics (MD) simulations validated against two-dimensional infrared (2D-IR) measurements of CO2 in an imidazolium-based ionic liquid have revealed new insights into the mechanism of CO2 solvation. The first solvation shell around CO2 has a distinctly quadrupolar structure, with strong negative charge density around the CO2 carbon atom and positive charge density near the CO2 oxygen atoms. When CO2 is modeled without atomic charges (thus removing its strong quadrupole moment), its solvation shell weakens and changes significantly into a structure that is similar to that of N2 in the same liquid. The solvation shell of CO2 evolves more quickly when its quadrupole is removed, and we find evidence that solvent cage dynamics is measured by 2D-IR spectroscopy. We also find that the solvent cage evolution of N2 is similar to that of CO2 with no atomic charges, implying that the weaker quadrupole of N2 is responsible for its higher diffusion and lower absorption in ionic liquids.

Decomposition of the Experimental Raman and Infrared Spectra of Acidic Water into Proton, Special Pair, and Counterion Contributions.

Textbooks describe excess protons in liquid water as hydronium (H3O+) ions, although their true structure remains lively debated. To address this question, we have combined Raman and infrared (IR) multivariate curve resolution spectroscopy with ab initio molecular dynamics and anharmonic vibrational spectroscopic calculations. Our results are used to resolve, for the first time, the vibrational spectra of hydrated protons and counterions and reveal that there is little ion-pairing below 2 M. Moreover, we find that isolated excess protons are strongly IR active and nearly Raman inactive (with vibrational frequencies of ∼1500 ± 500 cm-1), while flanking water OH vibrations are both IR and Raman active (with higher frequencies of ∼2500 ± 500 cm-1). The emerging picture is consistent with Georg Zundel's seminal work, as well as recent ultrafast dynamics studies, leading to the conclusion that protons in liquid water are primarily hydrated by two flanking water molecules, with a broad range of proton hydrogen bond lengths and asymmetries.

Modeling Carbon Dioxide Vibrational Frequencies in Ionic Liquids: I. Ab Initio Calculations.

This work elucidates the molecular binding mechanism of CO2 in [C4C1IM][PF6] ionic liquid (IL) and its interplay with the CO2 asymmetric stretch frequency ν3, and establishes computational protocols for the reliable construction of spectroscopic maps for simulating ultrafast 2D-IR data of CO2 solvated in ILs. While charge transfer drives the static frequency shift between different ionic liquids ( J. Chem. Phys. 2015 , 142 , 212425 ), we find here that electrostatic and Pauli repulsion effects dominate the dynamical frequency shift between different geometries sampled from the finite-temperature dynamics within a single ionic liquid. This finding is also surprising because dispersion interactions dominate the CO2-IL interaction energies, but are comparably constant across different geometries. An important practical consequence of this finding is that density functional theory is expected to be sufficiently accurate for constructing potential energy surfaces for CO2 in [C4C1IM][PF6], as needed for accurate anharmonic calculations to construct a reliable spectroscopic map. Similarly, we established appropriate computational and chemical models for treating the extended solvent environment. We found that a QM/MM treatment including at least 2 cation-ion pairs at the QM level and at least 32 pairs at the MM level is necessary to converge vibrational frequencies to within 1 cm-1. Using these insights, this work identifies a computational protocol as well as a chemical model necessary to construct accurate spectroscopic maps from first principles.

Modeling Carbon Dioxide Vibrational Frequencies in Ionic Liquids: II. Spectroscopic Map.

The primary challenge for connecting molecular dynamics (MD) simulations to linear and two-dimensional infrared measurements is the calculation of the vibrational frequency for the chromophore of interest. Computing the vibrational frequency at each time step of the simulation with a quantum mechanical method like density functional theory (DFT) is generally prohibitively expensive. One approach to circumnavigate this problem is the use of spectroscopic maps. Spectroscopic maps are empirical relationships that correlate the frequency of interest to properties of the surrounding solvent that are readily accessible in the MD simulation. Here, we develop a spectroscopic map for the asymmetric stretch of CO2 in the 1-butyl-3-methylimidazolium hexafluorophosphate ([C4C1im][PF6]) ionic liquid (IL). DFT is used to compute the vibrational frequency of 500 statistically independent CO2-[C4C1im][PF6] clusters extracted from an MD simulation. When the map was tested on 500 different CO2-[C4C1im][PF6] clusters, the correlation coefficient between the benchmark frequencies and the predicted frequencies was R = 0.94, and the root-mean-square error was 2.7 cm-1. The calculated distribution of frequencies also agrees well with experiment. The spectroscopic map required information about the CO2 angle, the electrostatics of the surrounding solvent, and the Lennard-Jones interaction between the CO2 and the IL. The contribution of each term in the map was investigated using symmetry-adapted perturbation theory calculations.