More Like this

1. Dataset: 1,000 QM/MM minimum energy pathway conformations for the deacylation reactions of GES-5/imipenem

   https://doi.org/10.5281/zenodo.6473977  

   Song, Zilin; Tao, Peng (January 2022, Zenodo)

   This dataset consists of 1,000 coordinate files (in the CHARMM psf/cor format) for the QM/MM minimum energy pathways of the deacylation reactions between a Class A beta-lactamases (GES-5) and the imipenem antibiotic molecules.</p> All pathway conformations were optimized at DFTB3/3OB-f/CHARMM36 level with 36 replicas.</p> All single point calculations and charge population analysis were done at B3LYP-D3/6-31+G(d,p)/CHARMM36 level.</p> 0.paths_ges_imi_d1.tar.gz: 500 pathway conformations for GES-5/IPM-Delta1 deacylation reactions.</li>0.paths_ges_imi_d2.tar.gz: 500 pathway conformations for GES-5/IPM-Delta1 deacylation reactions.</li>1.eners.zip: The single point replica energies along all GES-5/IPM pathways.</li>1.chrgs.zip: The NBO charges of the QM region of all replica conformations along all GES-5/IPM pathways.</li>2.datasets.zip: The Python codes to postprocess the molecular data and the featurized the NumPy arrays.</li>3.gnn.zip: The Python codes that implements the edge-conditioned graph convolutional NN to predict the deacylation barriers.</li>5.representative_conf.zip: The pathway conformations of all cluster centroids and an energetic representative (pathway id 22) pathway. Note: This file also serves as a peephole of how the pathway conformations from Reaction Path with Holonomic Constrains calculations looks like.</li>6.benchmark.zip: The benchmark calculations that validates the DFTB3/3OB-f/CHARMM36 against DFTB3/3OB/CHARMM36 and B3LYP/6-31G(d,p)/CHARMM36 level of theory on the energetic representative (pathway id 22) pathway conformations. </li></ul> 

   more » « less
2. UNRAVELING MOLECULAR MECHANISMS OF ANTIBIOTIC RESISTANCE THROUGH MULTISCALE SIMULATIONS AND EXPLAINABLE MACHINE LEARNING

   Song, Zilin (May 2022, SMU Scholar)

   Pathogen resistance to β-lactam antibiotics compromises effective treatments of superbug infections. One major source of β-lactam resistance is the bacterial production of β-lactamases, which could effectively hydrolyze β-lactam drugs. In this thesis, the hydrolysis of various β-lactam antibiotics by class A serine-based β-lactamases (ASβLs) were investigated using hybrid Quantum Mechanical / Molecular Mechanical (QM/MM) minimum energy pathway (MEP) calculations and explainable machine learning (ML) approaches. The TEM-1/benzylpenicillin acylation reaction with QM/MM chain-of-states reaction pathways was firstly revisited. I proposed two decomposition methods for energy contribution analysis based on perturbing ML regression models. Both methods were shown to be model implementation invariant and successfully bridged the discrepancies between two pioneering mechanistic studies. The Toho-1 ASβL acylations of ampicillin and cefalexin were then investigated. I reported that the acylation pathway selection can be ligand dependent: ampicillin could undergo acylation via Lys73 or Glu166 acting as the general base while cefalexin acylation is limited to Lys73 as the general base. An explainable artificial intelligence (XAI) method, the Boltzmann-weighted Cumulative Integrated Gradients (BCIG), was developed to explain the different acylation pathway viability found for ampicillin and cefalexin. Lastly, conformational factors determining the GES-5/imipenem deacylation activity was investigated using edge-conditioned convolutional graph-learning (GL) methods. Critical mechanistic insights were derived from perturbative response of the GL latent representations, which explained the different deacylation reactivity between the two imipenem pyrroline tautomer states and identified the orientation of the carbapenem 6α-hydroxyethyl as the key factor that impacts the deacylation barrier heights. In summary, my thesis focuses on bridging QM/MM chain-of-states reaction pathway calculations and explainable ML to derive essential mechanistic insights into β-lactam resistance driven by ASβLs. 

   more » « less
3. QM/MM modeling of class A β-lactamases reveals distinct acylation pathways for ampicillin and cefalexin

   https://doi.org/10.1039/d1ob01593a  

   Song, Zilin; Trozzi, Francesco; Palzkill, Timothy; Tao, Peng (November 2021, Organic & Biomolecular Chemistry)

   Efficient mechanism-based design of antibiotics that are not susceptible to β-lactamases is hindered by the lack of comprehensive knowledge on the energetic landscapes for the hydrolysis of various β-lactams. Herein, we adopted efficient quantum mechanics/molecular mechanics simulations to explore the acylation reaction catalyzed by CTX-M-44 (Toho-1) β-lactamase. We show that the catalytic pathways for β-lactam hydrolysis are correlated to substrate scaffolds: using Glu166 as the only general base for acylation is viable for ampicillin but prohibitive for cefalexin. The present computational workflow provides quantitative insights to facilitate the optimization of future β-lactam antibiotics. 

   more » « less
4. Mitigating RF Jamming Attacks at the Physical Layer with Machine Learning Dataset

   https://doi.org/10.5281/zenodo.6304194  

   Jacovic, Marko; Rivas, Xaime Rey (January 2022, Zenodo)

   {"Abstract":["Data files were used in support of the research paper titled \u201cMitigating RF Jamming Attacks at the Physical Layer with Machine Learning<\/em>" which has been submitted to the IET Communications journal.<\/p>\n\n---------------------------------------------------------------------------------------------<\/p>\n\nAll data was collected using the SDR implementation shown here: https://github.com/mainland/dragonradio/tree/iet-paper. Particularly for antenna state selection, the files developed for this paper are located in 'dragonradio/scripts/:'<\/p>\n\n'ModeSelect.py': class used to defined the antenna state selection algorithm<\/li>'standalone-radio.py': SDR implementation for normal radio operation with reconfigurable antenna<\/li>'standalone-radio-tuning.py': SDR implementation for hyperparameter tunning<\/li>'standalone-radio-onmi.py': SDR implementation for omnidirectional mode only<\/li><\/ul>\n\n---------------------------------------------------------------------------------------------<\/p>\n\nAuthors: Marko Jacovic, Xaime Rivas Rey, Geoffrey Mainland, Kapil R. Dandekar\nContact: krd26@drexel.edu<\/p>\n\n---------------------------------------------------------------------------------------------<\/p>\n\nTop-level directories and content will be described below. Detailed descriptions of experiments performed are provided in the paper.<\/p>\n\n---------------------------------------------------------------------------------------------<\/p>\n\nclassifier_training: files used for training classifiers that are integrated into SDR platform<\/p>\n\n'logs-8-18' directory contains OTA SDR collected log files for each jammer type and under normal operation (including congested and weaklink states)<\/li>'classTrain.py' is the main parser for training the classifiers<\/li>'trainedClassifiers' contains the output classifiers generated by 'classTrain.py'<\/li><\/ul>\n\npost_processing_classifier: contains logs of online classifier outputs and processing script<\/p>\n\n'class' directory contains .csv logs of each RTE and OTA experiment for each jamming and operation scenario<\/li>'classProcess.py' parses the log files and provides classification report and confusion matrix for each multi-class and binary classifiers for each observed scenario - found in 'results->classifier_performance'<\/li><\/ul>\n\npost_processing_mgen: contains MGEN receiver logs and parser<\/p>\n\n'configs' contains JSON files to be used with parser for each experiment<\/li>'mgenLogs' contains MGEN receiver logs for each OTA and RTE experiment described. Within each experiment logs are separated by 'mit' for mitigation used, 'nj' for no jammer, and 'noMit' for no mitigation technique used. File names take the form *_cj_* for constant jammer, *_pj_* for periodic jammer, *_rj_* for reactive jammer, and *_nj_* for no jammer. Performance figures are found in 'results->mitigation_performance'<\/li><\/ul>\n\nray_tracing_emulation: contains files related to Drexel area, Art Museum, and UAV Drexel area validation RTE studies.<\/p>\n\nDirectory contains detailed 'readme.txt' for understanding.<\/li>Please note: the processing files and data logs present in 'validation' folder were developed by Wolfe et al. and should be cited as such, unless explicitly stated differently. \n\tS. Wolfe, S. Begashaw, Y. Liu and K. R. Dandekar, "Adaptive Link Optimization for 802.11 UAV Uplink Using a Reconfigurable Antenna," MILCOM 2018 - 2018 IEEE Military Communications Conference (MILCOM), 2018, pp. 1-6, doi: 10.1109/MILCOM.2018.8599696.<\/li><\/ul>\n\t<\/li><\/ul>\n\nresults: contains results obtained from study<\/p>\n\n'classifier_performance' contains .txt files summarizing binary and multi-class performance of online SDR system. Files obtained using 'post_processing_classifier.'<\/li>'mitigation_performance' contains figures generated by 'post_processing_mgen.'<\/li>'validation' contains RTE and OTA performance comparison obtained by 'ray_tracing_emulation->validation->matlab->outdoor_hover_plots.m'<\/li><\/ul>\n\ntuning_parameter_study: contains the OTA log files for antenna state selection hyperparameter study<\/p>\n\n'dataCollect' contains a folder for each jammer considered in the study, and inside each folder there is a CSV file corresponding to a different configuration of the learning parameters of the reconfigurable antenna. The configuration selected was the one that performed the best across all these experiments and is described in the paper.<\/li>'data_summary.txt'this file contains the summaries from all the CSV files for convenience.<\/li><\/ul>"]} 

   more » « less
5. The NANOGrav Search for Signals from New Physics: MCMC chains

   https://doi.org/10.5281/zenodo.8083620  

   NANOGrav, The Collaboration (January 2023, Zenodo)

   {"Abstract":["MCMC chains for the GWB analyses performed in the paper "The NANOGrav 15 yr Data Set: Search for Signals from New Physics<\/em>". <\/p>\n\nThe data is provided in pickle format. Each file contains a NumPy array with the MCMC chain (with burn-in already removed), and a dictionary with the model parameters' names as keys and their priors as values. You can load them as<\/p>\n\nwith open ('path/to/file.pkl', 'rb') as pick:\n temp = pickle.load(pick)\n\n params = temp[0]\n chain = temp[1]<\/code>\n\nThe naming convention for the files is the following:<\/p>\n\nigw<\/strong>: inflationary Gravitational Waves (GWs)<\/li>sigw: scalar-induced GWs\n\tsigw_box<\/strong>: assumes a box-like feature in the primordial power spectrum.<\/li>sigw_delta<\/strong>: assumes a delta-like feature in the primordial power spectrum.<\/li>sigw_gauss<\/strong>: assumes a Gaussian peak feature in the primordial power spectrum.<\/li><\/ul>\n\t<\/li>pt: cosmological phase transitions\n\tpt_bubble<\/strong>: assumes that the dominant contribution to the GW productions comes from bubble collisions.<\/li>pt_sound<\/strong>: assumes that the dominant contribution to the GW productions comes from sound waves.<\/li><\/ul>\n\t<\/li>stable: stable cosmic strings\n\tstable-c<\/strong>: stable strings emitting GWs only in the form of GW bursts from cusps on closed loops.<\/li>stable-k<\/strong>: stable strings emitting GWs only in the form of GW bursts from kinks on closed loops.<\/li>stable<\/strong>-m<\/strong>: stable strings emitting monochromatic GW at the fundamental frequency.<\/li>stable-n<\/strong>: stable strings described by numerical simulations including GWs from cusps and kinks.<\/li><\/ul>\n\t<\/li>meta: metastable cosmic strings\n\tmeta<\/strong>-l<\/strong>: metastable strings with GW emission from loops only.<\/li>meta-ls<\/strong> metastable strings with GW emission from loops and segments.<\/li><\/ul>\n\t<\/li>super<\/strong>: cosmic superstrings.<\/li>dw: domain walls\n\tdw-sm<\/strong>: domain walls decaying into Standard Model particles.<\/li>dw-dr<\/strong>: domain walls decaying into dark radiation.<\/li><\/ul>\n\t<\/li><\/ul>\n\nFor each model, we provide four files. One for the run where the new-physics signal is assumed to be the only GWB source. One for the run where the new-physics signal is superimposed to the signal from Supermassive Black Hole Binaries (SMBHB), for these files "_bhb" will be appended to the model name. Then, for both these scenarios, in the "compare" folder we provide the files for the hypermodel runs that were used to derive the Bayes' factors.<\/p>\n\nIn addition to chains for the stochastic models, we also provide data for the two deterministic models considered in the paper (ULDM and DM substructures). For the ULDM model, the naming convention of the files is the following (all the ULDM signals are superimposed to the SMBHB signal, see the discussion in the paper for more details)<\/p>\n\nuldm_e<\/strong>: ULDM Earth signal.<\/li>uldm_p: ULDM pulsar signal\n\tuldm_p_cor<\/strong>: correlated limit<\/li>uldm_p_unc<\/strong>: uncorrelated limit<\/li><\/ul>\n\t<\/li>uldm_c: ULDM combined Earth + pulsar signal direct coupling \n\tuldm_c_cor<\/strong>: correlated limit<\/li>uldm_c_unc<\/strong>: uncorrelated limit<\/li><\/ul>\n\t<\/li>uldm_vecB: vector ULDM coupled to the baryon number\n\tuldm_vecB_cor:<\/strong> correlated limit<\/li>uldm_vecB_unc<\/strong>: uncorrelated limit <\/li><\/ul>\n\t<\/li>uldm_vecBL: vector ULDM coupled to B-L\n\tuldm_vecBL_cor:<\/strong> correlated limit<\/li>uldm_vecBL_unc<\/strong>: uncorrelated limit<\/li><\/ul>\n\t<\/li>uldm_c_grav: ULDM combined Earth + pulsar signal for gravitational-only coupling\n\tuldm_c_grav_cor: correlated limit\n\t\tuldm_c_cor_grav_low<\/strong>: low mass region  <\/li>uldm_c_cor_grav_mon<\/strong>: monopole region<\/li>uldm_c_cor_grav_low<\/strong>: high mass region<\/li><\/ul>\n\t\t<\/li>uldm_c_unc<\/strong>: uncorrelated limit\n\t\tuldm_c_unc_grav_low<\/strong>: low mass region  <\/li>uldm_c_unc_grav_mon<\/strong>: monopole region<\/li>uldm_c_unc_grav_low<\/strong>: high mass region<\/li><\/ul>\n\t\t<\/li><\/ul>\n\t<\/li><\/ul>\n\nFor the substructure (static) model, we provide the chain for the marginalized distribution (as for the ULDM signal, the substructure signal is always superimposed to the SMBHB signal)<\/p>"]} 

   more » « less