Journal: ACS Chemical Biology

Article Title: Machine Learning Uncovers Natural Product Modulators of the 5-Lipoxygenase Pathway and Facilitates the Elucidation of Their Biological Mechanisms

doi: 10.1021/acschembio.3c00725

Figure Lengend Snippet: Workflow of predicting natural product 5-LO inhibitors. Known small-molecule—5-LO interactions were cleaned up; the molecules were stratified into strong and weak/nonmodulators (cutoff IC 50 = 1 μM). Numerical structure descriptors were generated for each molecule. Three predictive classification models were built (gradient boosted trees, logistic regression, and naive Bayesian) based on the descriptors. The models were then used to predict the activities of 24,500 commercially available natural products and natural product derivatives. A subset of triple- and double-positive predictions were further investigated in the PubChem database. Eleven natural products with bioactivities that could be consistent with 5-LO inhibition were chosen for further testing. A similar approach was taken to predict 5-LO pathway inhibitors from the data set of FLAP activity modulators, as described in the text.

Article Snippet: Data set of 5-LO modulators used to build ML models, data set of commercially available natural products, data set of FLAP modulators used to build ML models, metrics of ML models used in this study, list of molecular fragments in 5-LO modulators and their frequencies in strong and weak modulator subsets, 5-LO modulator-derived ML model predictions on the natural product data set, similarity search of tested natural products in the 5-LO modulator data set, FLAP modulator-derived ML model predictions on the natural product data set, KNIME workflow used to build ML models, and KNIME workflow used to carry out 5-LO fragment analysis ( PDF ) 5-LO molecular data set, commercially available natural products, FLAP modular data set, ML model metrics, 5-LO modulator fragment analysis, predictions from the 5-LO data set, predictions from the FLAP data set, ML workflow, and fragment analysis workflow( ZIP )

Techniques: Generated, Inhibition, Activity Assay

Journal: Building and environment

Article Title: Weather Correlations to Calculate Infiltration Rates for U. S. Commercial Building Energy Models

doi: 10.1016/j.buildenv.2017.10.029

Figure Lengend Snippet: Comparing annual HVAC energy costs of EnergyPlus (original) and EnergyPlus (correlations) models

Article Snippet: The Large Office was not used in this study because the height of the EnergyPlus ASHRAE 90.1-2013 prototype building is 37.5 m, but was 47.5 m in the ASHRAE 90.1-2004 reference building [ 33 ].

Techniques:

Journal: Building and environment

Article Title: Weather Correlations to Calculate Infiltration Rates for U. S. Commercial Building Energy Models

doi: 10.1016/j.buildenv.2017.10.029

Figure Lengend Snippet: Annual predicted percentage increase in HVAC electrical and gas use for each building model and climate zone using EnergyPlus (correlations) model over using EnergyPlus (original) models (without air barriers)

Article Snippet: The Large Office was not used in this study because the height of the EnergyPlus ASHRAE 90.1-2013 prototype building is 37.5 m, but was 47.5 m in the ASHRAE 90.1-2004 reference building [ 33 ].

Techniques:

Journal: Building and environment

Article Title: Weather Correlations to Calculate Infiltration Rates for U. S. Commercial Building Energy Models

doi: 10.1016/j.buildenv.2017.10.029

Figure Lengend Snippet: Additional percentage savings in annual HVAC (a) electrical and (b) gas use for each building types and climate zone using EnergyPlus (correlations) models over using EnergyPlus (original) models after building envelope airtightening Note: Negative values mean a relative decrease in HVAC energy use after building envelope airtightening using EnergyPlus (correlations) models over using EnergyPlus (original) models

Article Snippet: The Large Office was not used in this study because the height of the EnergyPlus ASHRAE 90.1-2013 prototype building is 37.5 m, but was 47.5 m in the ASHRAE 90.1-2004 reference building [ 33 ].

Techniques: