Abstract:A key factor in computational drug design is the consistency and reliability with which intermolecular interactions between a wide variety of molecules can be described. Here we present a procedure to efficiently, reliably and automatically assign partial atomic charges to atoms based on known distributions. We formally introduce the molecular charge assignment problem, where the task is to select a charge from a set of candidate charges for every atom of a given query molecule. Charges are accompanied by a sc… Show more
“…Approach (2) includes charge equilibration methods (QEq), 34,35 statistical machine learning models, [36][37][38][39] and nearest-neighbor-like approaches based on the chemical element and bonding environment of the atom. [40][41][42][43] Generally, approach (1) produces a more accurate electrostatic potential in the pores of the MOF but incurs a computational cost orders of magnitude greater than the cost of approach (2). Thankfully, Nazarian et al 44 performed periodic DFT calculations to obtain the electron densities in ca.…”
Virtual screenings can accelerate and reduce the cost of discovering metal-organic frameworks (MOFs) for their applications in gas storage, separation, and sensing. In molecular simulations of gas adsorption/diffusion in MOFs, the adsorbate-MOF electrostatic interaction is typically modeled by placing partial point charges on the atoms of the MOF. For the virtual screening of large libraries of MOFs, it is critical to develop computationally inexpensive methods to assign atomic partial charges to MOFs that accurately reproduce the electrostatic potential in their pores. Herein, we design and train a message passing neural network (MPNN) to predict the atomic partial charges on MOFs under a charge neutral constraint. A set of ca. 2,250 MOFs labeled with high-fidelity partial charges, derived from periodic electronic structure calculations, serves as training examples. In an end-to-end manner, from charge-labeled crystal graphs representing MOFs, our MPNN machine-learns features of the local bonding environments of the atoms and learns to predict partial atomic charges from these features. Our trained MPNN assigns high-fidelity partial point charges to MOFs with orders of magnitude lower computational cost than electronic structure calculations. To enhance the accuracy of virtual screenings of large libraries of MOFs for their adsorption-based applications, we make our trained MPNN model and MPNN-charge-assigned computation-ready, experimental MOF structures publicly available. File list (2) download file view on ChemRxiv Message_passing_networks_to_learn_MOF_charges.pdf (3.62 MiB) download file view on ChemRxiv Message_passing_networks_to_learn_MOF_charges_SI.... (1.55 MiB)
“…Approach (2) includes charge equilibration methods (QEq), 34,35 statistical machine learning models, [36][37][38][39] and nearest-neighbor-like approaches based on the chemical element and bonding environment of the atom. [40][41][42][43] Generally, approach (1) produces a more accurate electrostatic potential in the pores of the MOF but incurs a computational cost orders of magnitude greater than the cost of approach (2). Thankfully, Nazarian et al 44 performed periodic DFT calculations to obtain the electron densities in ca.…”
Virtual screenings can accelerate and reduce the cost of discovering metal-organic frameworks (MOFs) for their applications in gas storage, separation, and sensing. In molecular simulations of gas adsorption/diffusion in MOFs, the adsorbate-MOF electrostatic interaction is typically modeled by placing partial point charges on the atoms of the MOF. For the virtual screening of large libraries of MOFs, it is critical to develop computationally inexpensive methods to assign atomic partial charges to MOFs that accurately reproduce the electrostatic potential in their pores. Herein, we design and train a message passing neural network (MPNN) to predict the atomic partial charges on MOFs under a charge neutral constraint. A set of ca. 2,250 MOFs labeled with high-fidelity partial charges, derived from periodic electronic structure calculations, serves as training examples. In an end-to-end manner, from charge-labeled crystal graphs representing MOFs, our MPNN machine-learns features of the local bonding environments of the atoms and learns to predict partial atomic charges from these features. Our trained MPNN assigns high-fidelity partial point charges to MOFs with orders of magnitude lower computational cost than electronic structure calculations. To enhance the accuracy of virtual screenings of large libraries of MOFs for their adsorption-based applications, we make our trained MPNN model and MPNN-charge-assigned computation-ready, experimental MOF structures publicly available. File list (2) download file view on ChemRxiv Message_passing_networks_to_learn_MOF_charges.pdf (3.62 MiB) download file view on ChemRxiv Message_passing_networks_to_learn_MOF_charges_SI.... (1.55 MiB)
“…6 . Conifer provides clonal tree based on bulk and single-cell data while clonal evolution in the original study is inferred based on single-cell exome and copy number data, and PhISCS has taken the matrix of single-cell sequencing data and the estimated noise rates of SCS experiment as its inputs (the clonal tree is reported in the study of Karpov et al [ 34 ]). Fig.…”
Background
Genetic heterogeneity of a cancer tumor that develops during clonal evolution is one of the reasons for cancer treatment failure, by increasing the chance of drug resistance. Clones are cell populations with different genotypes, resulting from differences in somatic mutations that occur and accumulate during cancer development. An appropriate approach for identifying clones is determining the variant allele frequency of mutations that occurred in the tumor. Although bulk sequencing data can be used to provide that information, the frequencies are not informative enough for identifying different clones with the same prevalence and their evolutionary relationships. On the other hand, single-cell sequencing data provides valuable information about branching events in the evolution of a cancerous tumor. However, the temporal order of mutations may be determined with ambiguities using only single-cell data, while variant allele frequencies from bulk sequencing data can provide beneficial information for inferring the temporal order of mutations with fewer ambiguities.
Result
In this study, a new method called Conifer (ClONal tree Inference For hEterogeneity of tumoR) is proposed which combines aggregated variant allele frequency from bulk sequencing data with branching event information from single-cell sequencing data to more accurately identify clones and their evolutionary relationships. It is proven that the accuracy of clone identification and clonal tree inference is increased by using Conifer compared to other existing methods on various sets of simulated data. In addition, it is discussed that the evolutionary tree provided by Conifer on real cancer data sets is highly consistent with information in both bulk and single-cell data.
Conclusions
In this study, we have provided an accurate and robust method to identify clones of tumor heterogeneity and their evolutionary history by combining single-cell and bulk sequencing data.
“…Meanwhile, the reorganization of electrons among atoms and the formation of molecules are complex and multifaceted processes, and their full description is only possible within the boundaries of quantum mechanics (QM) (Bristow et al, 2014;Sanvito, 2019). Density functional theory (DFT) is the most common choice for routine ground-state calculations; however, the number of valence electrons scaled cubically, increasing the computational costs significantly (Engler et al, 2019). It will therefore not be suitable, especially when one needs to sample extended size and time scales.…”
Section: Introductionmentioning
confidence: 99%
“…In recent years, many efforts have been directed to the efficient improvement of force fields. In particular, machine learning combined with molecular simulation has been verified by many groups to be effective to develop force field including inferring charges based on a set of reference molecules (Botu et al, 2016;Chen et al, 2018;Inokuchi et al, 2018;Engler et al, 2019;Hu et al, 2019;Roman et al, 2019;Sanvito, 2019;Unke and Meuwly, 2019;Ye et al, 2019). Among these, the random forest regression (RFR) method has been proven to be feasible for the prediction of atomic charge without expending much effort on parameter tuning or descriptor selection.…”
The random forest regression (RFR) model was introduced to predict the multiple spin state charges of a heme model, which is important for the molecular dynamic simulation of the spin crossover phenomenon. In this work, a multiple spin state structure data set with 39,368 structures of the simplified heme-oxygen binding model was built from the non-adiabatic dynamic simulation trajectories. The ESP charges of each atom were calculated and used as the real-valued response. The conformational adapted charge model (CAC) of three spin states was constructed by an RFR model using symmetry functions. The results show that our RFR model can effectively predict the on the fly atomic charges with the varying conformations as well as the atomic charge of different spin states in the same conformation, thus achieving the balance of accuracy and efficiency. The average mean absolute error of the predicted charges of each spin state is <0.02 e. The comparison studies on descriptors showed a maximum 0.06 e improvement in prediction of the charge of Fe 2+ by using 11 manually selected structural parameters. We hope that this model can not only provide variable parameters for developing the force field of the multi-spin state but also facilitate automation, thus enabling large-scale simulations of atomistic systems.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
