Copertina

ATMOMACCS: An interpretable molecular descriptor for machine learning predictions in atmospheric science

Fuente: Zenodo

Salvato in:

Dettagli Bibliografici
Autore principale:	Lind, Linus Emil Elias
Natura:	Recurso digital
Pubblicazione:	Zenodo 2025
Soggetti:	molecular descriptor molecular fingerprint kernel ridge regression atmospheric organic compound aerosol
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

_version_	1866902177232977920
author	Lind, Linus Emil Elias
author_facet	Lind, Linus Emil Elias
contents	<h2>ATMOMACCS-KRR</h2> <p>ATMOMACCS-KRR is a research project exploring molecular descriptor construction and Kernel Ridge Regression (KRR) modeling for predicting thermodynamic and physicochemical properties of atmospheric organic molecules.</p> <p>The project introduces the ATMOMACCS descriptors (versions 1–5), which extend the traditional MACCS fingerprint to better represent structural features relevant in atmospheric chemistry.</p> <h2>Project outline</h2> <p>Main Script:</p> <ul> <li> <p><code>main.py</code><br>End-to-end pipeline: generates descriptors, trains a KRR model, and produces predictions.</p> </li> </ul> <p>Descriptor Construction:</p> <ul> <li> <p>ATMOMACCS v1–v4:<br><code>ATMOMACCS.py</code><br>Implements versions v1–v4 of ATMOMACCS descriptors.</p> </li> <li> <p>ATMOMACCS v5:<br><code>ATMOMACCS_no_binary.py</code><br>Implements version v5 of ATMOMACCS.</p> </li> </ul> <p>Descriptor Generation Scripts:</p> <ul> <li> <p><code>generate_ATMOMACCS.py</code><br>Generate ATMOMACCS descriptors for a dataset.</p> </li> <li> <p><code>generate_MACCS.py</code><br>Generate standard MACCS keys as baseline descriptors.</p> </li> <li> <p><code>generate_optimal_topfp.py</code><br>Generate optimized topological fingerprints.</p> </li> <li> <p><code>generate_topological.py</code><br>Generate topological fingerprints.</p> </li> </ul> <p>Results Processing:</p> <ul> <li> <p><code>process_results.py</code><br>Aggregates results across multiple random seeds and produces mean and standard deviation CSV summaries for each descriptor/target combination.</p> </li> </ul> <h2>Getting Started</h2> <p>Example run with <code>main.py</code>:</p> <p><code>python src/main.py -v 4 -d data/Wang -t log_p_sat.txt -ds Wang -s 2435</code></p> <p>This will:</p> <ul> <li> <p>Generate descriptors (ATMOMACCS v4 in this case)</p> </li> <li> <p>Train a KRR model on the chosen dataset and target</p> </li> <li> <p>Save predictions and performance metrics to the results folder</p> </li> </ul> <p>Requirements</p> <p>This project was developed and tested with Python 3.12.0</p> <blockquote> <p>Dependencies:</p> <ul> <li> <p><code>matplotlib==3.8.0</code></p> </li> <li> <p><code>pandas==2.1.1</code></p> </li> <li> <p><code>rdkit==2023.9.1</code></p> </li> <li> <p><code>scikit-learn==1.3.2</code></p> </li> <li> <p><code>scipy==1.11.3</code></p> </li> </ul> </blockquote> <h2>Licensing and Data Sources</h2> <p>The ATMOMACCS-KRR code (scripts, descriptor generation, and processing workflows) is licensed under the<strong> </strong>Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0). This repository redistributes and processes several benchmark datasets. However, the <strong>datasets included in this repository</strong> (Ferraz-Caetano, GeckoQ, Wang, Li) are <strong>redistributed under their original licenses</strong>. Modifications and preprocessing applied to the datasets are documented in this repository. Note that the <strong>original data licenses must be respected</strong> if the datasets are reused, redistributed, or adapted. Please see the text section below for full attribution and licensing information for each dataset.</p> <p>Ferraz-Caetano:</p> <ul> <li> <p>License: MIT License <a href="https://github.com/jfcaetano/VOC-EnthVapML/blob/main/LICENSE">https://github.com/jfcaetano/VOC-EnthVapML/blob/main/LICENSE</a></p> </li> <li> <p>Original data file: VOC-Database.csv</p> </li> <li> <p>Source: <a href="https://github.com/jfcaetano/VOC-EnthVapML/tree/main/Database" target="_new" rel="noopener">https://github.com/jfcaetano/VOC-EnthVapML/tree/main/Database</a></p> </li> <li> <p>Publication: José Ferraz-Caetano, Filipe Teixeira, M. Natália D.S. Cordeiro, Data-driven, explainable machine learning model for predicting volatile organic compounds’ standard vaporization enthalpy, Chemosphere, Volume 359, 142257 (2024)<br>DOI: <a href="https://doi.org/10.1016/j.chemosphere.2024.142257" target="_new" rel="noopener">https://doi.org/10.1016/j.chemosphere.2024.142257</a></p> </li> </ul> <p>GeckoQ:</p> <ul> <li> <p>License: Creative Commons Attribution 4.0 International (CC BY 4.0)</p> </li> <li> <p>Original data file: Dataframe.csv</p> </li> <li> <p>Source: <a href="https://doi.org/10.23729/022475cc-e527-41a9-bbc0-0113923cf04c" target="_new" rel="noopener">https://doi.org/10.23729/022475cc-e527-41a9-bbc0-0113923cf04c</a></p> </li> <li> <p>Publication: Vitus Besel, Milica Todorović, Theo Kurtén, Patrick Rinke & Hanna Vehkamäki, Atomic structures, conformers and thermodynamic properties of 32k atmospheric molecules, Scientific Data 10, 450 (2023)<br>DOI: <a href="https://doi.org/10.1038/s41597-023-02366-x" target="_new" rel="noopener">https://doi.org/10.1038/s41597-023-02366-x</a></p> </li> </ul> <p>Wang:</p> <ul> <li> <p>License: Creative Commons Attribution 3.0 Unported (CC BY 3.0)</p> </li> <li> <p>Original data file: Supplementary data (acp-17-7529-2017-supplement.zip)</p> </li> <li> <p>Source: <a href="https://acp.copernicus.org/articles/17/7529/2017/acp-17-7529-2017-supplement.zip" target="_new" rel="noopener">https://acp.copernicus.org/articles/17/7529/2017/acp-17-7529-2017-supplement.zip</a></p> </li> <li> <p>Publication: Chen Wang, Tiange Yuan, Stephen A. Wood, Kai-Uwe Goss, Jingyi Li, Qi Ying, and Frank Wania, Uncertain Henry's law constants compromise equilibrium partitioning calculations of atmospheric oxidation products, Atmos. Chem. Phys., 17, 7529–7540 (2017)<br>DOI: <a href="https://doi.org/10.5194/acp-17-7529-2017" target="_new" rel="noopener">https://doi.org/10.5194/acp-17-7529-2017</a></p> </li> <li> <p>Modifications: Documented in data_preprocessing.ipynb (formatting and restructuring of files)</p> </li> </ul> <p>Li:</p> <ul> <li> <p>License: Creative Commons Attribution 4.0 International (CC BY 4.0)</p> </li> <li> <p>Original data file: Li et al. OA viscosity_Table S2.xls</p> </li> <li> <p>Source: <a href="https://doi.org/10.5194/acp-20-8103-2020-supplement" target="_new" rel="noopener">https://doi.org/10.5194/acp-20-8103-2020-supplement</a></p> </li> <li> <p>Publication: Ying Li, Douglas A. Day, Harald Stark, Jose L. Jimenez, and Manabu Shiraiwa, Predictions of the glass transition temperature and viscosity of organic aerosols from volatility distributions, Atmos. Chem. Phys., 20, 8103–8122 (2020)<br>DOI: <a href="https://doi.org/10.5194/acp-20-8103-2020" target="_new" rel="noopener">https://doi.org/10.5194/acp-20-8103-2020</a></p> </li> <li> <p>Modifications: Documented in CAS_to_smiles.ipynb (formatting and restructuring of files)</p> </li> </ul>
format	Recurso digital
id	zenodo_https___doi_org_10_5281_zenodo_17231684
institution	Zenodo
language
publishDate	2025
publisher	Zenodo
record_format	zenodo
spellingShingle	ATMOMACCS: An interpretable molecular descriptor for machine learning predictions in atmospheric science Lind, Linus Emil Elias molecular descriptor molecular fingerprint kernel ridge regression atmospheric organic compound aerosol <h2>ATMOMACCS-KRR</h2> <p>ATMOMACCS-KRR is a research project exploring molecular descriptor construction and Kernel Ridge Regression (KRR) modeling for predicting thermodynamic and physicochemical properties of atmospheric organic molecules.</p> <p>The project introduces the ATMOMACCS descriptors (versions 1–5), which extend the traditional MACCS fingerprint to better represent structural features relevant in atmospheric chemistry.</p> <h2>Project outline</h2> <p>Main Script:</p> <ul> <li> <p><code>main.py</code><br>End-to-end pipeline: generates descriptors, trains a KRR model, and produces predictions.</p> </li> </ul> <p>Descriptor Construction:</p> <ul> <li> <p>ATMOMACCS v1–v4:<br><code>ATMOMACCS.py</code><br>Implements versions v1–v4 of ATMOMACCS descriptors.</p> </li> <li> <p>ATMOMACCS v5:<br><code>ATMOMACCS_no_binary.py</code><br>Implements version v5 of ATMOMACCS.</p> </li> </ul> <p>Descriptor Generation Scripts:</p> <ul> <li> <p><code>generate_ATMOMACCS.py</code><br>Generate ATMOMACCS descriptors for a dataset.</p> </li> <li> <p><code>generate_MACCS.py</code><br>Generate standard MACCS keys as baseline descriptors.</p> </li> <li> <p><code>generate_optimal_topfp.py</code><br>Generate optimized topological fingerprints.</p> </li> <li> <p><code>generate_topological.py</code><br>Generate topological fingerprints.</p> </li> </ul> <p>Results Processing:</p> <ul> <li> <p><code>process_results.py</code><br>Aggregates results across multiple random seeds and produces mean and standard deviation CSV summaries for each descriptor/target combination.</p> </li> </ul> <h2>Getting Started</h2> <p>Example run with <code>main.py</code>:</p> <p><code>python src/main.py -v 4 -d data/Wang -t log_p_sat.txt -ds Wang -s 2435</code></p> <p>This will:</p> <ul> <li> <p>Generate descriptors (ATMOMACCS v4 in this case)</p> </li> <li> <p>Train a KRR model on the chosen dataset and target</p> </li> <li> <p>Save predictions and performance metrics to the results folder</p> </li> </ul> <p>Requirements</p> <p>This project was developed and tested with Python 3.12.0</p> <blockquote> <p>Dependencies:</p> <ul> <li> <p><code>matplotlib==3.8.0</code></p> </li> <li> <p><code>pandas==2.1.1</code></p> </li> <li> <p><code>rdkit==2023.9.1</code></p> </li> <li> <p><code>scikit-learn==1.3.2</code></p> </li> <li> <p><code>scipy==1.11.3</code></p> </li> </ul> </blockquote> <h2>Licensing and Data Sources</h2> <p>The ATMOMACCS-KRR code (scripts, descriptor generation, and processing workflows) is licensed under the<strong> </strong>Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0). This repository redistributes and processes several benchmark datasets. However, the <strong>datasets included in this repository</strong> (Ferraz-Caetano, GeckoQ, Wang, Li) are <strong>redistributed under their original licenses</strong>. Modifications and preprocessing applied to the datasets are documented in this repository. Note that the <strong>original data licenses must be respected</strong> if the datasets are reused, redistributed, or adapted. Please see the text section below for full attribution and licensing information for each dataset.</p> <p>Ferraz-Caetano:</p> <ul> <li> <p>License: MIT License <a href="https://github.com/jfcaetano/VOC-EnthVapML/blob/main/LICENSE">https://github.com/jfcaetano/VOC-EnthVapML/blob/main/LICENSE</a></p> </li> <li> <p>Original data file: VOC-Database.csv</p> </li> <li> <p>Source: <a href="https://github.com/jfcaetano/VOC-EnthVapML/tree/main/Database" target="_new" rel="noopener">https://github.com/jfcaetano/VOC-EnthVapML/tree/main/Database</a></p> </li> <li> <p>Publication: José Ferraz-Caetano, Filipe Teixeira, M. Natália D.S. Cordeiro, Data-driven, explainable machine learning model for predicting volatile organic compounds’ standard vaporization enthalpy, Chemosphere, Volume 359, 142257 (2024)<br>DOI: <a href="https://doi.org/10.1016/j.chemosphere.2024.142257" target="_new" rel="noopener">https://doi.org/10.1016/j.chemosphere.2024.142257</a></p> </li> </ul> <p>GeckoQ:</p> <ul> <li> <p>License: Creative Commons Attribution 4.0 International (CC BY 4.0)</p> </li> <li> <p>Original data file: Dataframe.csv</p> </li> <li> <p>Source: <a href="https://doi.org/10.23729/022475cc-e527-41a9-bbc0-0113923cf04c" target="_new" rel="noopener">https://doi.org/10.23729/022475cc-e527-41a9-bbc0-0113923cf04c</a></p> </li> <li> <p>Publication: Vitus Besel, Milica Todorović, Theo Kurtén, Patrick Rinke & Hanna Vehkamäki, Atomic structures, conformers and thermodynamic properties of 32k atmospheric molecules, Scientific Data 10, 450 (2023)<br>DOI: <a href="https://doi.org/10.1038/s41597-023-02366-x" target="_new" rel="noopener">https://doi.org/10.1038/s41597-023-02366-x</a></p> </li> </ul> <p>Wang:</p> <ul> <li> <p>License: Creative Commons Attribution 3.0 Unported (CC BY 3.0)</p> </li> <li> <p>Original data file: Supplementary data (acp-17-7529-2017-supplement.zip)</p> </li> <li> <p>Source: <a href="https://acp.copernicus.org/articles/17/7529/2017/acp-17-7529-2017-supplement.zip" target="_new" rel="noopener">https://acp.copernicus.org/articles/17/7529/2017/acp-17-7529-2017-supplement.zip</a></p> </li> <li> <p>Publication: Chen Wang, Tiange Yuan, Stephen A. Wood, Kai-Uwe Goss, Jingyi Li, Qi Ying, and Frank Wania, Uncertain Henry's law constants compromise equilibrium partitioning calculations of atmospheric oxidation products, Atmos. Chem. Phys., 17, 7529–7540 (2017)<br>DOI: <a href="https://doi.org/10.5194/acp-17-7529-2017" target="_new" rel="noopener">https://doi.org/10.5194/acp-17-7529-2017</a></p> </li> <li> <p>Modifications: Documented in data_preprocessing.ipynb (formatting and restructuring of files)</p> </li> </ul> <p>Li:</p> <ul> <li> <p>License: Creative Commons Attribution 4.0 International (CC BY 4.0)</p> </li> <li> <p>Original data file: Li et al. OA viscosity_Table S2.xls</p> </li> <li> <p>Source: <a href="https://doi.org/10.5194/acp-20-8103-2020-supplement" target="_new" rel="noopener">https://doi.org/10.5194/acp-20-8103-2020-supplement</a></p> </li> <li> <p>Publication: Ying Li, Douglas A. Day, Harald Stark, Jose L. Jimenez, and Manabu Shiraiwa, Predictions of the glass transition temperature and viscosity of organic aerosols from volatility distributions, Atmos. Chem. Phys., 20, 8103–8122 (2020)<br>DOI: <a href="https://doi.org/10.5194/acp-20-8103-2020" target="_new" rel="noopener">https://doi.org/10.5194/acp-20-8103-2020</a></p> </li> <li> <p>Modifications: Documented in CAS_to_smiles.ipynb (formatting and restructuring of files)</p> </li> </ul>
title	ATMOMACCS: An interpretable molecular descriptor for machine learning predictions in atmospheric science
topic	molecular descriptor molecular fingerprint kernel ridge regression atmospheric organic compound aerosol
url	https://doi.org/10.5281/zenodo.17231684