Search for: All records

Random forest is considered as one of the most successful machine learning algorithms, which has been widely used to construct microbiome-based predictive models. However, its use as a statistical testing method has not been explored. In this study, we propose “Random Forest Test” (RFtest), a global (community-level) test based on random forest for high-dimensional and phylogenetically structured microbiome data. RFtest is a permutation test using the generalization error of random forest as the test statistic. Our simulations demonstrate that RFtest has controlled type I error rates, that its power is superior to competing methods for phylogenetically clustered signals, and that it is robust to outliers and adaptive to interaction effects and non-linear associations. Finally, we apply RFtest to two real microbiome datasets to ascertain whether microbial communities are associated or not with the outcome variables. 

Abstract As one of the least understood aerosol processes, nucleation can be a dominant source of atmospheric aerosols. Sulfuric acid (SA)-amine binary nucleation with dimethylamine (DMA) has been recognized as a governing mechanism in the polluted continental boundary layer. Here we demonstrate the importance of trimethylamine (TMA) for nucleation in the complex atmosphere and propose a molecular-level SA-DMA-TMA ternary nucleation mechanism as an improvement upon the conventional binary mechanism. Using the proposed mechanism, we could connect the gaseous amines to the SA-amine cluster signals measured in the atmosphere of urban Beijing. Results show that TMA can accelerate the SA-DMA-based new particle formation in Beijing by 50–100%. Considering the global abundance of TMA and DMA, our findings imply comparable importance of TMA and DMA to nucleation in the polluted continental boundary layer, with probably higher contributions from TMA in polluted rural environments and future urban environments with controlled DMA emissions. 

Investigating the adsorption of organic pollutants onto boron nitride nanosheets is crucial for designing novel boron nitride adsorbents so as to remove pollutants from the environment. In this study, we performed density functional theory (DFT) computations to investigate the adsorption of 28 aromatic compounds onto boron nitride nanosheets, and developed four quantitative structure–activity relationship (QSAR) models for predicting the logarithm of the adsorption equilibrium constant (log  K ) values of organic pollutants adsorbed onto boron nitride nanosheets in both gaseous and aqueous environments. The DFT-predicted adsorption energies showed that boron nitride nanosheets exhibit stronger adsorption capability than graphene. Our QSAR analyses revealed that van der Waals interactions play dominant roles in gaseous adsorption, while van der Waals and hydrophobic interactions are the main driving forces in aqueous adsorption. This work demonstrates that in silico QSAR models can serve as efficient tools for high-throughput prediction of log  K values for organic pollutants adsorbed onto boron nitride nanomaterials. 

Predicting adsorption of organic pollutants onto graphene nanomaterials is not only useful for exploring their potential adsorbent applications, but also helpful for better understanding their fate and risks in aquatic environments. Herein molecular dynamics (MD) simulations and theoretical linear solvation energy relationships (TLSERs) were employed to construct prediction models for adsorption of neutral organic pollutants onto graphene and graphene oxides. The MD simulations for adsorption of 43 aromatic compounds onto graphene and diverse models of graphene oxides with various functional groups (hydroxyl, epoxy and carbonyl) demonstrate that graphene has a stronger affinity for the aromatic compounds than graphene oxides. The hydroxyl and carbonyl groups of graphene oxides were found to form hydrogen bonds with the aromatic adsorbates, while epoxy groups did not. TLSER models were developed for predicting the adsorption equilibrium coefficients ( K ) onto graphene and graphene oxide nanosheets. In the graphene prediction model, H-donating ability ( ε α ) and dispersion/hydrophobic interactions ( V ) have significant effects on log  K values, while in the graphene oxide model, ε α is the most influential factor on log  K values. The models provide in silico approaches for predicting adsorption affinities onto graphenic nanomaterials.