Find research datasets worth reusing
Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.
13,064
datasets available to search
ShareScore release 0.9.0
Dataset results
13,064 results for “predictions”
Robust and efficient reranking in crystal structure prediction: a data driven method for real-life molecules
<p>The content of this repository accompanies the publication "Robust and efficient reranking in crystal structure prediction: a data driven method for real-life molecules" and contains the complete dataset produced for the fentanyl CSP. </p> <p>Three types of datasets are present : Generation, ML-Reranker and GRACE. <br>Generation refers to all the molecular crystal structures that GRACE has generated using the tailor made force field. It's the starting pool of the reranking exercise, and it contains all the structures that will be selected by the reranking processes.<br>ML-Reranker refers to the data generated by the algorithm proposed in our manuscript. The configurations and energies are obtained by selecting from structures from the generation pool and relaxing their coordinates.<br>GRACE dataset contains the configurations which a user obtains at the end of a standard GRACE reranking procedure. Since GRACE follows differet convergence and minimization critera, the structures obtained in this dataset can differ (non-substantially) from the equivalents found in the ML-Reranker. </p> <p>*.data : contains the indices, energy of the crystal structure (kcal/mol) and, in case of the ml-reranker dataset, the indices mapping the obtained landscape to their generating pool.</p> <p>*.xyz : contains ASE formatted, extended-xyz list of structures corresponding to each exercise.</p> <p>Authors:</p> <p>Andrea Anelli, Hanno Dietrich, Philipp Ectors, Frank Stowasser, Tristan Bereau, Marcus Neumann, Joost van den Ende</p>
Code and data from: Towards a more dynamic metabolic theory of ecology to predict climate change effects on biological systems
<p>This repository contains data and code used to produce Figures 1, 2, and S1 in <em>Towards a more dynamic metabolic theory of ecology to predict climate change effects on biological systems.</em></p> <p>The file "empirical_mte_database.csv" contains a database of peer-reviewed articles pulled from a Web of Science search of papers that empirically tested the temperature dependence predictions of the metabolic theory of ecology (Brown et al. 2004) between 2004 and 2024. </p> <p>The .zip file contains a jupyter notebook, julia project toml file and a README in order to reproduce the simulations illustrating how different temperature dynamics can lead to different inferred thermal performance curves in populations.</p> <p> </p>
Reactzyme: A Benchmark for Enzyme-Reaction Prediction
<h3>Official dataset of <span>Reactzyme -</span> <span>Reactzyme: A Benchmark for</span><span>Enzyme-Reaction Prediction.</span></h3> <p><span>Our study utilizes a comprehensive dataset compiled from the SwissProt and Rhea databases. SwissProt, a curated subset of the UniProt database, has been selected for its high-quality, human-derived functional annotations of protein sequences. This section of UniProt is particularly valuable for its expert-reviewed entries, which ensure reliable and accurate functional data, making it ideal for our analysis. Rhea is employed for its precise mapping from enzymes to specific catalyzed functions, offering detailed descriptions of biochemical reactions. </span></p> <p><span>The SwissProt and Rhea dataset are downloaded on January 8, 2024, and includes data entries up to this date, providing the most recent and comprehensive data available for our study. We selectively exclude water molecules and unspecific functional groups that could mask the true molecular structures. Conversely, we remove metal ions, gas molecules, and other small molecules because of their potential to bind to proteins, a characteristic that presents a valuable learning feature for our model. To this end, the total dataset comprises 178,463 positive enzyme-reaction pairs, including 178,327 unique enzymes and 7,726 unique reactions. </span></p>
Neural network prediction of strong lensing systems with domain adaptation and uncertainty quantification
<p>This project combines the emerging field of Domain Adaptation with Uncertainty Quantification, working towards applying machine learning to real scientific datasets with limited labelled data. For this project, simulated images of strong gravitational lenses are used as source and target dataset, and the Einstein radius θ E and its uncertainty are determined through regression.</p> <p>Applying machine learning in science domains such as astronomy is difficult. With models trained on simulated data being applied to real data, models frequently underperform - simulations cannot perfectlty capture the true complexity of real data. Enter domain adaptation (DA). The DA techniques used in this work use Maximum Mean Discrepancy (MMD) Loss to train a network to being embeddings of labelled "source" data gravitational lenses in line with unlabeled "target" gravitational lenses. With source and target datasets made similar, training on source datasets can be used with greater fidelity on target datasets.</p> <p>Scientific analysis requires an estimate of uncertainty on measurements. We adopt an approach known as mean-variance estimation, which seeks to estimate the variance and control regression by minimizing the beta negative log-likelihood loss. To our knowledge, this is the first time that domain adaptation and uncertainty quantification are being combined, especially for regression on an astrophysical dataset.</p>
Individualistic reward-seeking strategies that predict response to nicotine emerge among isogenic male mice living in a micro-society
<p>This repository contains data and basic code for analysis and modeling used in the following paper: Fayad et al. ""Individualistic reward-seeking strategies that predict response to nicotine emerge among isogenic male mice living in a micro-society", published on BioRxiv (https://www.biorxiv.org/content/10.1101/2023.10.28.564522v1)</p> <p> </p>
Dataset: Asymptotic-state prediction for fast flavor transformation in neutron star mergers
<p>These data accompany the paper "Asymptotic-state predictions for fast flavor transformations in neutron star mergers" by S. Richers, J. Froustey, S. Ghosh, F. Foucart, and J. Gomez. The data are sorted into three directories:</p> <p>In the datasets, spacetime components are generally ordered as:<br>0 = x<br>1 = y<br>2 = z<br>3 = t</p> <p>Neutrino/antineutrino indices are ordered as:<br>0 = neutrino<br>1 = antineutrino</p> <p>Flavor indices are ordered as:<br>0 = e<br>1 = mu<br>2 = tau</p> <p>#==========#<br># Emu_data #<br>#==========#<br>Contains four files, each containing the initial and final results of simulations of neutrino quantum kinetics.</p> <p>M1NuLib-2016_5ms_rl0.h5: extracted from refinement level 0 of a 1.2Msun-1.2Msun NS merger simulation at 5ms after merger (https://doi.org/10.1103/PhysRevD.94.123016).<br>M1NuLib_3ms_rl1.h5: extracted from refinement level 1 of a 1.3Msun-1.4Msun NS merger at 3ms after merger (https://arxiv.org/abs/2407.15989)<br>M1NuLib_7ms_rl1.h5: extracted from refinement level 1 of a 1.3Msun-1.4Msun NS merger at 7ms after merger (https://arxiv.org/abs/2407.15989)<br>random: randomly generated initial conditions as described in the manuscript.</p> <p>Contents:</p> <p>F4_initial(1|ccm):<br>[simulation index, 4 spacetime components, 2 neutrino/antineutrino, 3 flavors]<br>The initial number density four-flux of each species in units of 1/cm^3.</p> <p>F4_final(1|ccm):<br>[simulation index, 4 spacetime components, 2 neutrino/antineutrino, 3 flavors]<br>The final number density four-flux of each species in units of 1/cm^3, averaged in time following the procedure indicated in the accompanying paper.</p> <p>F4_final_stddev(1|ccm):<br>[simulation index, 4 spacetime components, 2 neutrino/antineutrino, 3 flavors]<br>Not used in the manuscript. The standard deviation from the distribution of F4 that the final answer is averaged over.</p> <p>directorynames:<br>[simulation index]<br>A string for each simulation index indicating the directory the calculation corresponds to. Used in debugging.</p> <p>growthRate(1|s):<br>[simulation index]<br>Not used in the manuscript. Measured growth rate of the flavor off-diagonal components of the number density in s^-1. N_offdiag ~ e^(w t), where w is the growth rate and t is the time.</p> <p>nf:<br>scalar<br>Assumed number of flavors. Should be 3 for all datasets included here.</p> <p>xplot:<br>[simulation index, time index]<br>t/t_saturation. Used in creating Figure 2 of the manuscript.</p> <p>y0plot:<br>[simulation index, time index]<br>N_ee(t) / N_ee(0). Used in creating Figure 2 of the manuscript.</p> <p>y1plot:<br>[simulation index, time index]<br>N_offdiag_mag(t) / Ntot. Used in creating Figure 2 of the manuscript.</p> <p><br>#===========#<br># SpEC_data #<br>#===========#<br>Contains snapshots of grid quantities extracted from neutron star merger simulations. The file M1NuLib_2016_5ms_rl0.h5 contains data from the 1.2Msun-1.2Msun simulation of https://doi.org/10.1103/PhysRevD.94.123016. All others contain data from the 1.3Msun-1.4Msun M1-NuLib simulation of https://doi.org/10.48550/arXiv.2407.15989. The (3ms, 5ms, 7ms) part of the filename indicates how much time after merger the snapshot was taken. The (rl0, rl1, rl2, rl3) part of the filename indicates which refinement level the data are extracted from. NaNs in the data indicate that that cell was covered by a finer refinement level in the simulation. Radiation quantities are Lorentz transformed into an orthonormal tetrad comoving with the background fluid, and are rotated such that the net ELN flux is in the z direction. Each dataset contains:</p> <p>J_{e,a,x}(erg|ccm) - energy density of {electron neutrinos, electron anti-neutrinos, heavy lepton neutrinos}, where "x" contains the sum of all four heavy species, in units if erg/cm^3. Indexed by spatial position of the grid cell: [i,j,k]</p> <p>fn_{e,a,x}(1|ccm) - number flux of {electron neutrinos, electron anti-neutrinos, heavy lepton neutrinos}, where "x" contains the sum of all four heavy species, in units if 1/cm^3. Indexed by direction and spatial position of the grid cell: [xyz, i,j,k]</p> <p>minerbo_Z{e,a,x} - the Z parameter of the maximum entropy closure at each point for {electron neutrinos, electron anti-neutrinos, heavy lepton neutrinos}. Dimensionless.</p> <p>n_{e,a,x}(1|ccm) - number diensity of {electron neutrinos, electron anti-neutrinos, heavy lepton neutrinos} in units of 1/cm^3. Indexed by spatial position of the grid cell: [i,j,k]</p> <p>{x,y,z}(cm) - coordinates of the centers of each grid cell in units of cm. Indexed by spatial position of the grid cell: [i,j,k]</p> <p>rho(g|ccm) - background comoving matter density in units of g/cm^3. Indexed by spatial position of the grid cell: [i,j,k]</p> <p>T(MeV) - background comoving matter temperature in units of MeV. Indexed by spatial position of the grid cell: [i,j,k]</p> <p>Ye - background electron fraction (dimensionless). Indexed by spatial position of the grid cell: [i,j,k]</p> <p>fluxfac_{e,a,x} - flux factor of {electron neutrinos, electron anti-neutrinos, heavy lepton neutrinos}. Dimensionless.</p> <p>crossing_discriminant - a crossing exists if this number is larger than 0. Computed using the following, based on Equation 16 in https://doi.org/10.1103/PhysRevD.106.083005. Indexed by spatial position of the grid cell: [i,j,k]<br>a = gamma**2 + alpha**2<br>b = -2. * gamma * eta<br>c = eta**2 - alpha**2<br>discriminant = (b**2 - 4.*a*c) / (2.*a)**2<br>hasCrossing = (discriminant >= 0)</p> <p>deltaCrossingAngle - width of the ELN crossing based on Equation 16 in https://doi.org/10.1103/PhysRevD.106.083005 in units of radians. Indexed by spatial position of the grid cell: [i,j,k]</p> <p>{nue,anue}_absrate(1|s) - absorption rate for {electron neutrinos, electron antineutrinos} in units if 1/s. Indexed by spatial position of the grid cell: [i,j,k]</p> <p>g{xx,yy,zz} - diagonal components of the metric tensor (dimensionless). Indexed by spatial position of the grid cell: [i,j,k]</p> <p><br>#===========#<br># ML_models #<br>#===========#<br>Contains the ML models used in the accompanying paper. In addition, there is an example script that uses the Rhea code to generate the relevant databases. Running the script requires the Rhea directory to be in the Python path. The Rhea code contains example Python and C++ code to use a trained model.</p> <p>Expected Rhea code:<br>Snapshot at 10.5281/zenodo.13675320<br>https://github.com/srichers/Rhea commit 68b7dc69d8fa33c7f5de1d8bec0790496f282d80</p> <p> </p>
Dataset from Makino H. and Suhaimi A. Distributed representations of temporally accumulated reward prediction errors in the mouse cortex.
<p>Dataset from the paper:</p> <p>Makino H. and Suhaimi A. Distributed representations of temporally accumulated reward prediction errors in the mouse cortex.</p> <p>Each variable is described in Description.pdf.</p> <p>Analysis code is available at <a href="https://github.com/HiroshiMakinoLaboratory/RPEAccumulation" target="_blank" rel="noopener">https://github.com/HiroshiMakinoLaboratory/RewardPredictionErrorAccumulation</a>.</p>
The effect of uncertainty in humidity and model parameters on the prediction of contrail energy forcing
<p>Previous work has shown that while the net effect of aircraft condensation trails (contrails) on the<br>climate is warming, the exact magnitude of the energy forcing per meter of contrail remains uncertain.<br>In this paper, we explore the skill of a Lagrangian contrail model (CoCiP) in identifying flight<br>segments with high contrail energy forcing. We find that skill is greater than climatological<br>predictions alone, even accounting for uncertainty in weather fields and model parameters.</p> <p>We estimate the uncertainty in weather by using the ensemble ERA5 weather reanalysis from the European<br>Centre for Medium-Range Weather Forecasts (ECMWF) as Monte Carlo inputs to CoCiP. We unbias and correct<br>under-dispersion on the ERA5 humidity data by forcing a match to the distribution of in situ humidity<br>measurements taken at cruising altitude. We set aside CoCiP energy forcing estimates calculated using<br>one of the ensemble members as a proxy for ground truth, and report the skill of CoCiP in identifying<br>segments with large positive proxy energy forcing. We further estimate the uncertainty in the model<br>parameters in CoCiP by performing Monte Carlo simulations with CoCiP model parameters drawn from<br>uncertainty distributions consistent with the literature.</p> <p>When CoCiP outputs are averaged over seasons to form climatological predictions, the skill in<br>predicting the proxy is 44%, while the skill of per-flight CoCiP outputs is 84%. If these results carry<br>over to the true (unknown) contrail EF, they indicate that per-flight energy forcing predictions can<br>reduce the number of potential contrail avoidance route adjustments by 2x, hence reducing both the cost<br>and fuel impact of contrail avoidance.</p>
How to select predictive models for decision making or causal inference? Experiments data
<p>This is the full result data for the experiments of the paper : Doutreligne, M., & Varoquaux, G. (2023). How to select predictive models for decision making or causal inference?, https://hal.science/hal-03946902. <br><br>The code repository is : https://github.com/soda-inria/caussim/tree/main</p> <p>The files in this dataset are the one for the most computationnally costly experiments. There is one folder for each of the four datasets used in the paper. Then, one folder for each of the experimental setup. The files required for the main figure (Fig.3) of the paper are the one labelled #fig3 in the following descriptions.</p> <p>Details on the files : </p> <p>.<br>├── acic_2016_save<br>│ ├── acic_2016__nuisance_non_linear__candidates_hist_gradient_boosting__dgp_1-77__rs_1-5<br>│ │ └── run_logs.csv: results for the experiment with non linear models for both the nuisances and the candidates<br>│ ├── acic_2016__nuisance_non_linear__candidates_ridge__dgp_1-77__rs_1-10<br>│ │ └── run_logs.csv: results for the experiment with non linear models for the nuisances and linear models for the candidates<br>│ └── acic_2016__stacked_regressor__dgp_1-77__seed_1-10<br>│ └── run_logs.csv: results for the experiment with stacked models (linear and non linear) for the nuisances and non linear models for the candidates #fig3<br>├── acic_2018_save<br>│ └── acic_2018__nuisance_non_linear__candidates_hist_gradient_boosting__first_uid_432<br>│ └── run_logs.csv results for the experiment with stacked models (linear and non linear) for the nuisances models and non linear models for the candidates #fig3<br>├── caussim_save<br>│ ├── caussim__linear_regressor__test_size_5000__n_datasets_1000<br>│ │ ├── run_logs.csv: results for the experiment with stacked models for the nuisances models and linear models for the candidates <br>│ │ └── simu.yaml: configuration file of the experiment<br>│ ├── caussim__nuisance_non_linear__candidates_ridge__overlap_01-247_join_nuisance_train_set<br>│ │ └── run_logs.csv: results for the experiment with non linear models for the nuisances and linear models for the candidates, joined sets for the nuisances and the candidates<br>│ ├── caussim__nuisance_non_linear__candidates_ridge__overlap_01-247_separated_nuisance_train_set<br>│ │ └── run_logs.csv: results for the experiment with non linear models for the nuisances and linear models for the candidates, separated sets for the nuisances and the candidates<br>│ └── caussim__stacked_regressor__test_size_5000__n_datasets_1000<br>│ ├── run_logs.csv: results for the experiment with stacked models (linear and non linear) for the nuisances and linear models for the candidates #fig3<br>│ └── simu.yaml: configuration file of the experiment<br>└── twins_save<br> └── twins__stacked_regressor__rs_1-10__overlap_0.1-3<br> └── run_logs.csv: results for the experiment with stacked models (linear and non linear) for the nuisances and non linear models for the candidates #fig3</p>
Chronic Apical Periodontitis Risk Prediction and Assessment
<p><strong>Workflow and Process:</strong></p> <ol> <li> <p><strong>Sample Selection and Preparation:</strong></p> <ul> <li><strong>Objective:</strong> Select relevant CAP-related samples from the UK Biobank ensuring quality and representativeness.</li> <li><strong>Tasks:</strong> Integrate personal health information, blood biochemical indicators, and genetic data to construct feature sets.</li> </ul> <p><strong>Related Files/Directories:</strong></p> <ul> <li><code>S1_sample_selection.csv</code>: CSV file with selected samples.</li> <li><code>feature_info</code>: Directory containing information about features used.</li> </ul> </li> <li> <p><strong>Data Integration and Feature Engineering:</strong></p> <ul> <li><strong>Objective:</strong> Combine genetic and electronic medical record data to build comprehensive feature sets.</li> <li><strong>Tasks:</strong> Process and integrate various data types to prepare for model development.</li> </ul> <p><strong>Related Files/Directories:</strong></p> <ul> <li><code>data_integration</code>: Directory containing scripts and data files for integration.</li> </ul> </li> <li> <p><strong>Machine Learning Model Development:</strong></p> <ul> <li><strong>Objective:</strong> Develop and train various machine learning models, including random forest, support vector machine, logistic regression, and adaptive boosting, using the integrated data.</li> <li><strong>Tasks:</strong> Implement and train models to predict CAP risk based on the selected features.</li> </ul> <p><strong>Related Files/Directories:</strong></p> <ul> <li><code>code</code>: Directory containing code files for machine learning model development.</li> <li><code>models</code>: Directory with trained model files.</li> </ul> </li> <li> <p><strong>Model Performance Evaluation:</strong></p> <ul> <li><strong>Objective:</strong> Assess the performance of the machine learning models and compare their predictive capabilities.</li> <li><strong>Tasks:</strong> Evaluate models using metrics such as area under the curve (AUC) and sensitivity.</li> </ul> <p><strong>Related Files/Directories:</strong></p> <ul> <li><code>S2_traintest_id.RData</code>: RData file with training and testing identifiers.</li> <li><code>S4_model_performance.RData</code>: RData file with model performance results.</li> <li><code>S5_model_res.zip</code>: Compressed file containing detailed model results</li> </ul> </li> </ol>
Divergence time and environmental similarity predict the strength of morphological convergence in stick and leaf insects
<p>This uploads contains the datasets, phylogenetic tree and associated R code used to generate the results reported in the article: "Divergence time and environmental similarity predict the strength of morphological convergence in stick and leaf insects" published in Proceedings of the National Academy of Sciences USA (2024).<br>A detailed explanation of datasetS1 can be found in the supplementary data of the article. </p>
Out-of-Distribution Evaluation of Active Learning Pipelines for Molecular Property Prediction Dataset
<p>Data for the paper "Out-of-Distribution Evaluation of Active Learning Pipelines for Molecular Property Prediction"</p>
Mapping the relative accuracy of cross-ancestry prediction
<p>GWAS for six traits for SNPs from the UK Biobank arrays set, six traits for SNPs from the HapMap set, and three traits for SNPs from the ARIC study set.</p>
Supplementary Data associated with the article "LukProt: A database of eukaryotic predicted proteins designed for investigations of animal origins"
<p>Supplementary Data associated with the article "LukProt: A database of eukaryotic predicted proteins designed for investigations of animal origins" - BUSCO example.</p> <p> </p> <p>National Science Centre of Poland is acknowledged for funding of the project 2020/36/C/NZ8/00081, "The role of glycosylation in the emergence of animal multicellularity", which enabled the creation of this dataset.</p> <p> </p> <p> </p>
Datasets and Code: Revealing the Role of Redox Reaction Selectivity and Mass Transfer in Current–Voltage Predictions for Ensembles of Photocatalysts
<ul> <li>Raw datasets (.mat and .fig files) and codes (.mlx and .m files) used in our manuscript of the same title. </li> <li>Figure numbers correspond with the figure numbers in the corresponding manuscript. <ul> <li>Figure 4: Effects of kinetic parameters on Solar-to-chemical (STC) efficiencies and reaction selectivity</li> <li>Figure 5: Solar-to-chemical (STC) efficiencies for a model incorporating competing undesired redox reactions implemented for different redox shuttle pairs</li> <li>Figure 7: Solar-to-chemical efficiencies for an ensemble of light absorbers</li> <li>Figure 8: Maximum solar-to-chemical (STC) efficiencies and corresponding number of light absorbers as a function of asymmetry factors in limiting current density for redox shuttle reduction</li> <li>Figure 9: Solar-to-chemical efficiencies for an increasing number of light absorbers for different total absorptance values (99%, 75%, 50%).</li> <li>Figure 10: Qualitative comparisons between experimental measurements and model predictions for a photocatalytic suspension reactor</li> </ul> </li> <li>The main piece of the code developed is provided as an interactive .mlx file; not all subfunction calls within the main code is included, and can be shared upon reasonable request via email from the lead (luisab@umich.edu) and the corresponding authors (rbchan@umich.edu) of this paper. </li> </ul> <p> </p>
Observed data, predictions and uncertainty associated with the updated distribution of clay minerals in the World Ocean
<p>Sparase observed data for various clay mineral species are .csv file format.</p> <p>Predictions and uncertainty for four seafloor clay mineral species (relative percentages, Kaolinite, Illite, Smectite, Chlorite) are generated via geospatial machine learning (GML). Methodology is outlined in "The updated distribution of clay mineral in the World Ocean"<span>. These files are in net-CDF file format. Files are cell-centered. Further, latitudes and longitudes of grid cells are denoted in the variables of the .nc files.</span></p>
Yields, Cannabinoids Quantification, and Predictive Programming Codes Using Machine Learning for Non-Psychoactive Cannabis Flowers and Extracts (Cannabis sativa L.) Cultivated in Ecuador.
<p>This publication presents data from various extraction methods, including maceration, Soxhlet, and supercritical fluids, performed on different cannabis flower varieties (Cannabis sativa L.) under varying operating conditions. We quantified the amounts of CBD, THC, CBG, and CBN in the extracts produced by each method using High-Performance Liquid Chromatography (HPLC). Using this data, we developed a machine learning algorithm in RStudio to make predictions and determine the best conditions and yields for each extraction method. The analysis focuses on different varieties of non-psychoactive cannabis cultivated in Ecuador at altitudes over 2,450 m.a.s.l.</p>
LoGoFunc predictions
<p>Predictions indicating variants resulting in pathogenic gain- or loss-of-function or neutral variants for canonical missense variants in the human genome. Variant coordinates correspond to the GRCh38/hg38 genome assembly.</p> <p>Columns "LoGoFunc_neutral", "LoGoFunc_GOF", "LoGoFunc_LOF" correspond to the predicted probability that a variant is neutral, results in pathogenic gain-of-function, or results in pathogenic loss-of-function respectively. Higher scores indicating higher probability as predicted by the model.</p> <p>Available for academic use only. For commercial use contact the corresponding author.</p>
Global soil water content (volumetric m³/m³) for 10kPa, 33kPa and 1500kPa suctions predicted at 4 depths (0, 30, 60, and 100 cm) at 1km resolution
<p>Volumetric soil water content (m³/m³) at 10 kPa, 33 kPa, and 1500 kPa suctions was predicted at four depths (0, 30, 60, and 100 cm) with a spatial resolution of 1 km. The maps of van Genuchten (vG) parameters from Gupta et al. (2022) were used to calculate the soil water content at these pressures, following Equation 1 from Gupta et al. (2022).</p> <p>References:</p> <ol> <li>Gupta, Surya, Papritz, Andreas, Lehmann, Peter, Hengl, Tomislav, Bonetti, Sara, & Or, Dani. (2022). Global maps of soil water characteristics parameters developed using the random forest in a Covariate-based GeoTransfer Functions (CoGTF) framework at 1 km resolution [Data set]. https://doi.org/10.5281/zenodo.6343570</li> </ol>
Fast and Accurate Prediction of Tautomer Ratios in Aqueous Solution via a Siamese Neural Network
<p>Tautomerization plays a critical role in many chemical and biological processes, impacting the molecular stability, reactivity, biological activity, and ADME-Tox properties. Many drug-like molecules exist in multiple tautomeric states in aqueous solutions and complicating drug discovery. Predicting these tautomeric ratios and identifying the predominant species rapidly and accurately is crucial for computational drug discovery. In this study, we introduce sPhysNet-Taut, a deep learning model fine-tuned with experimental data leveraging the Siamese network. This model predicts tautomer ratios in aqueous solution using MMFF94-optimized geometries directly. On an experimental test set, sPhysNet-Taut surpasses all other methods, achieving state-of-the-art performance with an RMSE of 1.9 kcal/mol on the 100-tautomers set and an RMSE of 1.0 kcal/mol on the SAMPL2 challenge, and providing the best ranking power for tautomer pairs. Additionally, our results demonstrate that fine-tuning on experimental data significantly improves model performance compared to training from scratch. This work not only provides a useful deep learning model for predicting tautomer ratios, but also provides a protocol for modeling pairwise data. To facilitate user-friendliness, we developed a readily accessible tool to predict stable tautomeric states in aqueous solutions, enumerating all possible tautomeric states and ranking them using the sPhysNet-Taut model.</p>
ScienceDex guides
Understand access before you commit
These curated guides explain access requirements, typical timelines, costs, and reuse considerations for widely used research datasets.
Allen Brain Atlas
Allen Brain Atlas is an Allen Institute collection of brain map atlases, datasets, APIs, and analysis tools covering mouse, human, and non-human primate brain resources.
Annotated Behaviour and Observability Dataset (ABODe)
ABODe is a University of Edinburgh DataShare dataset for behavior classification in group-housed mice using home-cage video, identities, bounding boxes, ground-plate positions, and annotator labels.
DANDI Archive for NWB datasets
DANDI is a BRAIN Initiative archive for publishing and sharing neurophysiology data, including electrophysiology, optophysiology, and behavioral data packaged as NWB and related standards.
International Brain Laboratory public data
The International Brain Laboratory public data releases expose standardized mouse decision-making experiments, including Neuropixels recordings, widefield calcium imaging, behavior, and session metadata accessed through the ONE API.
OpenNeuro
OpenNeuro is a free, open platform for sharing neuroimaging datasets, with public search, dataset pages, and download paths for web, S3, DataLad, and the OpenNeuro CLI.