Early Career Perspectives from Large Pharma, Software, and Start-up Companies.

We are often approached by PhD students and postdocs who wonder: What are the differences between jobs for computational chemists across different industries? This Perspective aims to answer this question by comparing our personal experiences as early career scientists at a large pharmaceutical company (large pharma), a software vendor (software), and a biotech start-up (start-up) in the format of a written Q&A panel discussion. To begin, we introduce ourselves by answering questions about our backgrounds and current positions, including comparisons of our responsibilities and the culture of the companies where we work. In the next section, we focus on the beginning of our careers, discussing the skills we needed for our first industry positions and what we learned early on. Finally, we address questions about the future of our careers including potential growth, security, and what we wished we had known earlier. We conclude by comparing and contrasting our industries, including how the size and purpose of these companies have affected our experiences.

Development and Benchmarking of Open Force Field v1.0.0-the Parsley Small-Molecule Force Field.

We present a methodology for defining and optimizing a general force field for classical molecular simulations, and we describe its use to derive the Open Force Field 1.0.0 small-molecule force field, codenamed Parsley. Rather than using traditional atom typing, our approach is built on the SMIRKS-native Open Force Field (SMIRNOFF) parameter assignment formalism, which handles increases in the diversity and specificity of the force field definition without needlessly increasing the complexity of the specification. Parameters are optimized with the ForceBalance tool, based on reference quantum chemical data that include torsion potential energy profiles, optimized gas-phase structures, and vibrational frequencies. These quantum reference data are computed and are maintained with QCArchive, an open-source and freely available distributed computing and database software ecosystem. In this initial application of the method, we present essentially a full optimization of all valence parameters and report tests of the resulting force field against compounds and data types outside the training set. These tests show improvements in optimized geometries and conformational energetics and demonstrate that Parsley's accuracy for liquid properties is similar to that of other general force fields, as is accuracy on binding free energies. We find that this initial Parsley force field affords accuracy similar to that of other general force fields when used to calculate relative binding free energies spanning 199 protein-ligand systems. Additionally, the resulting infrastructure allows us to rapidly optimize an entirely new force field with minimal human intervention.

Improving small molecule force fields by identifying and characterizing small molecules with inconsistent parameters.

Many molecular simulation methods use force fields to help model and simulate molecules and their behavior in various environments. Force fields are sets of functions and parameters used to calculate the potential energy of a chemical system as a function of the atomic coordinates. Despite the widespread use of force fields, their inadequacies are often thought to contribute to systematic errors in molecular simulations. Furthermore, different force fields tend to give varying results on the same systems with the same simulation settings. Here, we present a pipeline for comparing the geometries of small molecule conformers. We aimed to identify molecules or chemistries that are particularly informative for future force field development because they display inconsistencies between force fields. We applied our pipeline to a subset of the eMolecules database, and highlighted molecules that appear to be parameterized inconsistently across different force fields. We then identified over-represented functional groups in these molecule sets. The molecules and moieties identified by this pipeline may be particularly helpful for future force field parameterization.

Toward Learned Chemical Perception of Force Field Typing Rules.

Molecular mechanics force fields define how the energy and forces in a molecular system are computed from its atomic positions, thus enabling the study of such systems through computational methods like molecular dynamics and Monte Carlo simulations. Despite progress toward automated force field parametrization, considerable human expertise is required to develop or extend force fields. In particular, human input has long been required to define atom types, which encode chemically unique environments that determine which parameters will be assigned. However, relying on humans to establish atom types is suboptimal. Human-created atom types are often developed without statistical justification, leading to over- or under-fitting of data. Human-created types are also difficult to extend in a systematic and consistent manner when new chemistries must be modeled or new data becomes available. Finally, human effort is not scalable when force fields must be generated for new (bio)polymers, compound classes, or materials. To remedy these deficiencies, our long-term goal is to replace human specification of atom types with an automated approach, based on rigorous statistics and driven by experimental and/or quantum chemical reference data. In this work, we describe novel methods that automate the discovery of appropriate chemical perception: SMARTY allows for the creation of atom types, while SMIRKY goes further by automating the creation of fragment (nonbonded, bonds, angles, and torsions) types. These approaches enable the creation of move sets in atom or fragment type space, which are used within a Monte Carlo optimization approach. We demonstrate the power of these new methods by automating the rediscovery of human defined atom types (SMARTY) or fragment types (SMIRKY) in existing small molecule force fields. We assess these approaches using several molecular data sets, including one which covers a diverse subset of the DrugBank database.

SAMPL6 challenge results from [Formula: see text] predictions based on a general Gaussian process model.

A variety of fields would benefit from accurate [Formula: see text] predictions, especially drug design due to the effect a change in ionization state can have on a molecule's physiochemical properties. Participants in the recent SAMPL6 blind challenge were asked to submit predictions for microscopic and macroscopic [Formula: see text]s of 24 drug like small molecules. We recently built a general model for predicting [Formula: see text]s using a Gaussian process regression trained using physical and chemical features of each ionizable group. Our pipeline takes a molecular graph and uses the OpenEye Toolkits to calculate features describing the removal of a proton. These features are fed into a Scikit-learn Gaussian process to predict microscopic [Formula: see text]s which are then used to analytically determine macroscopic [Formula: see text]s. Our Gaussian process is trained on a set of 2700 macroscopic [Formula: see text]s from monoprotic and select diprotic molecules. Here, we share our results for microscopic and macroscopic predictions in the SAMPL6 challenge. Overall, we ranked in the middle of the pack compared to other participants, but our fairly good agreement with experiment is still promising considering the challenge molecules are chemically diverse and often polyprotic while our training set is predominately monoprotic. Of particular importance to us when building this model was to include an uncertainty estimate based on the chemistry of the molecule that would reflect the likely accuracy of our prediction. Our model reports large uncertainties for the molecules that appear to have chemistry outside our domain of applicability, along with good agreement in quantile-quantile plots, indicating it can predict its own accuracy. The challenge highlighted a variety of means to improve our model, including adding more polyprotic molecules to our training set and more carefully considering what functional groups we do or do not identify as ionizable.

Escaping Atom Types in Force Fields Using Direct Chemical Perception.

Traditional approaches to specifying a molecular mechanics force field encode all the information needed to assign force field parameters to a given molecule into a discrete set of atom types. This is equivalent to a representation consisting of a molecular graph comprising a set of vertices, which represent atoms labeled by atom type, and unlabeled edges, which represent chemical bonds. Bond stretch, angle bend, and dihedral parameters are then assigned by looking up bonded pairs, triplets, and quartets of atom types in parameter tables to assign valence terms and using the atom types themselves to assign nonbonded parameters. This approach, which we call indirect chemical perception because it operates on the intermediate graph of atom-typed nodes, creates a number of technical problems. For example, atom types must be sufficiently complex to encode all necessary information about the molecular environment, making it difficult to extend force fields encoded this way. Atom typing also results in a proliferation of redundant parameters applied to chemically equivalent classes of valence terms, needlessly increasing force field complexity. Here, we describe a new approach to assigning force field parameters via direct chemical perception. Rather than working through the intermediary of the atom-typed graph, direct chemical perception operates directly on the unmodified chemical graph of the molecule to assign parameters. In particular, parameters are assigned to each type of force field term (e.g., bond stretch, angle bend, torsion, and Lennard-Jones) based on standard chemical substructure queries implemented via the industry-standard SMARTS chemical perception language, using SMIRKS extensions that permit labeling of specific atoms within a chemical pattern. We use this to implement a new force field format, called the SMIRKS Native Open Force Field (SMIRNOFF) format. We demonstrate the power and generality of this approach using examples of specific molecules that pose problems for indirect chemical perception and construct and validate a minimalist yet very general force field, SMIRNOFF99Frosst. We find that a parameter definition file only ∼300 lines long provides coverage of all but <0.02% of a 5 million molecule drug-like test set. Despite its simplicity, the accuracy of SMIRNOFF99Frosst for small molecule hydration free energies and selected properties of pure organic liquids is similar to that of the General Amber Force Field, whose specification requires thousands of parameters. This force field provides a starting point for further optimization and refitting work to follow.

Blind prediction of cyclohexane-water distribution coefficients from the SAMPL5 challenge.

In the recent SAMPL5 challenge, participants submitted predictions for cyclohexane/water distribution coefficients for a set of 53 small molecules. Distribution coefficients (log D) replace the hydration free energies that were a central part of the past five SAMPL challenges. A wide variety of computational methods were represented by the 76 submissions from 18 participating groups. Here, we analyze submissions by a variety of error metrics and provide details for a number of reference calculations we performed. As in the SAMPL4 challenge, we assessed the ability of participants to evaluate not just their statistical uncertainty, but their model uncertainty-how well they can predict the magnitude of their model or force field error for specific predictions. Unfortunately, this remains an area where prediction and analysis need improvement. In SAMPL4 the top performing submissions achieved a root-mean-squared error (RMSE) around 1.5 kcal/mol. If we anticipate accuracy in log D predictions to be similar to the hydration free energy predictions in SAMPL4, the expected error here would be around 1.54 log units. Only a few submissions had an RMSE below 2.5 log units in their predicted log D values. However, distribution coefficients introduced complexities not present in past SAMPL challenges, including tautomer enumeration, that are likely to be important in predicting biomolecular properties of interest to drug discovery, therefore some decrease in accuracy would be expected. Overall, the SAMPL5 distribution coefficient challenge provided great insight into the importance of modeling a variety of physical effects. We believe these types of measurements will be a promising source of data for future blind challenges, especially in view of the relatively straightforward nature of the experiments and the level of insight provided.

Calculating Partition Coefficients of Small Molecules in Octanol/Water and Cyclohexane/Water.

Partition coefficients describe how a solute is distributed between two immiscible solvents. They are used in drug design as a measure of a solute's hydrophobicity and a proxy for its membrane permeability. We calculate partition coefficients from transfer free energies using molecular dynamics simulations in explicit solvent. Setup is done by our new Solvation Toolkit which automates the process of creating input files for any combination of solutes and solvents for many popular molecular dynamics software packages. We calculate partition coefficients between octanol/water and cyclohexane/water with the Generalized AMBER Force Field (GAFF) and the Dielectric Corrected GAFF (GAFF-DC). With similar methods in the past we found a root-mean-squared error (RMSE) of 6.3 kJ/mol in hydration free energies which would correspond to an error of around 1.6 log units in partition coefficients if solvation free energies in both solvents were estimated with comparable accuracy. Here we find an overall RMSE of about 1.2 log units with both force fields. Results from GAFF and GAFF-DC seem to exhibit systematic biases in opposite directions for calculated cyclohexane/water partition coefficients.

Investigating vibrational relaxation in cyanide-bridged transition metal mixed-valence complexes using two-dimensional infrared and infrared pump-probe spectroscopies.

Using polarization-selective two-dimensional infrared (2D IR) and infrared pump-probe spectroscopies, we study vibrational relaxation of the four cyanide stretching (νCN) vibrations found in [(NH3)5Ru(III)NCFe(II)(CN)5](-) (FeRu) dissolved in D2O or formamide and [(NC)5Fe(II)CNPt(IV)(NH3)4NCFe(II)(CN)5](4-) (FePtFe) dissolved in D2O. These cyanide-bridged transition metal complexes serve as models for understanding the role high frequency vibrational modes play in metal-to-metal charge transfers over a bridging ligand. However, there is currently little information about vibrational relaxation and dephasing dynamics of the anharmonically coupled νCN modes in the electronic ground state of these complexes. IR pump-probe experiments reveal that the vibrational lifetimes of the νCN modes are ∼2 times faster when FeRu is dissolved in D2O versus formamide. They also reveal that the vibrational lifetimes of the νCN modes of FePtFe in D2O are almost four times as long as for FeRu in D2O. Combined with mode-specific relaxation dynamics measured from the 2D IR experiments, the IR pump-probe experiments also reveal that intramolecular vibrational relaxation is occurring in all three systems on ∼1 ps timescale. Center line slope dynamics, which have been shown to be a measure of the frequency-frequency correlation function, reveal that the radial, axial, and trans νCN modes exhibit a ∼3 ps timescale for frequency fluctuations. This timescale is attributed to the forming and breaking of hydrogen bonds between each mode and the solvent. The results presented here along with our previous work on FeRu and FePtFe reveal a picture of coupled anharmonic νCN modes where the spectral diffusion and vibrational relaxation dynamics depend on the spatial localization of the mode on the molecular complex and its specific interaction with the solvent.

Investigating vibrational anharmonic couplings in cyanide-bridged transition metal mixed valence complexes using two-dimensional infrared spectroscopy.

Using polarization-selective two-dimensional infrared (2D IR) spectroscopy, we measure anharmonic couplings and angles between the transition dipole moments of the four cyanide stretching (νCN) vibrations found in [(NH3)5Ru(III)NCFe(II)(CN)5](-) (FeRu) dissolved in D2O and formamide and [(NC)5Fe(II)CNPt(IV)(NH3)4NCFe(II)(CN)5](4-) (FePtFe) dissolved in D2O. These cyanide-bridged transition metal complexes serve as model systems for studying the role of high frequency vibrational modes in ultrafast photoinduced charge transfer reactions. Here, we focus on the spectroscopy of the νCN modes in the electronic ground state. The FTIR spectra of the νCN modes of the bimetallic and trimetallic systems are strikingly different in terms of frequencies, amplitudes, and lineshapes. The experimental 2D IR spectra of FeRu and FePtFe and their fits reveal a set of weakly coupled anharmonic νCN modes. The vibrational mode anharmonicities of the individual νCN modes range from 14 to 28 cm(-1). The mixed-mode anharmonicities range from 2 to 14 cm(-1). In general, the bridging νCN mode is most weakly coupled to the radial νCN mode, which involves the terminal CN ligands. Measurement of the relative transition dipole moments of the four νCN modes reveal that the FeRu molecule is almost linear in solution when dissolved in formamide, but it assumes a bent geometry when dissolved in D2O. The νCN modes are modelled as bilinearly coupled anharmonic oscillators with an average coupling constant of 6 cm(-1). This study elucidates the role of the solvent in modulating the molecular geometry and the anharmonic vibrational couplings between the νCN modes in cyanide-bridged transition metal mixed valence complexes.

Simulating Ru L3-edge X-ray absorption spectroscopy with time-dependent density functional theory: model complexes and electron localization in mixed-valence metal dimers.

Ruthenium L3-edge X-ray absorption (XA) spectroscopy probes unoccupied 4d orbitals of the metal atom and is increasingly being used to investigate the local electronic structure in ground and excited electronic states of Ru complexes. The simultaneous development of computational tools for simulating Ru L3-edge spectra is crucial for interpreting the spectral features at a molecular level. This study demonstrates that time-dependent density functional theory (TDDFT) is a viable and predictive tool for simulating ruthenium L3-edge XA spectroscopy. We systematically investigate the effects of exchange correlation functional and implicit and explicit solvent interactions on a series of Ru(II) and Ru(III) complexes in their ground and electronic excited states. The TDDFT simulations reproduce all of the experimentally observed features in Ru L3-edge XA spectra within the experimental resolution (0.4 eV). Our simulations identify ligand-specific charge transfer features in complicated Ru L3-edge spectra of [Ru(CN)6](4-) and Ru(II) polypyridyl complexes illustrating the advantage of using TDDFT in complex systems. We conclude that the B3LYP functional most accurately predicts the transition energies of charge transfer features in these systems. We use our TDDFT approach to simulate experimental Ru L3-edge XA spectra of transition metal mixed-valence dimers of the form [(NC)5M(II)-CN-Ru(III)(NH3)5](-) (where M = Fe or Ru) dissolved in water. Our study determines the spectral signatures of electron delocalization in Ru L3-edge XA spectra. We find that the inclusion of explicit solvent molecules is necessary for reproducing the spectral features and the experimentally determined valencies in these mixed-valence complexes. This study validates the use of TDDFT for simulating Ru 2p excitations using popular quantum chemistry codes and providing a powerful interpretive tool for equilibrium and ultrafast Ru L3-edge XA spectroscopy.