Metabolomics holds the promise as a new technology to diagnose highly heterogeneous diseases. Conventionally, metabolomics data analysis for diagnosis is done using various statistical and machine learning based classification methods. However, it remains unknown if deep neural network, a class of increasingly popular machine learning methods, is suitable to classify metabolomics data. Here we use a cohort of 271 breast cancer tissues, 204 positive estrogen receptor (ER+), and 67 negative estrogen receptor (ER−) to test the accuracies of feed-forward networks, a deep learning (DL) framework, as well as six widely used machine learning models, namely random forest (RF), support vector machines (SVM), recursive partitioning and regression trees (RPART), linear discriminant analysis (LDA), prediction analysis for microarrays (PAM), and generalized boosted models (GBM). DL framework has the highest area under the curve (AUC) of 0.93 in classifying ER+/ER– patients, compared to the other six machine learning algorithms. Furthermore, the biological interpretation of the first hidden layer reveals eight commonly enriched significant metabolomics pathways (adjusted P-value <0.05) that cannot be discovered by other machine learning methods. Among them, protein digestion and absorption and ATP-binding cassette (ABC) transporters pathways are also confirmed in integrated analysis between metabolomics and gene expression data in these samples. In summary, deep learning method shows advantages for metabolomics based breast cancer ER status classification, with both the highest prediction accuracy (AUC = 0.93) and better revelation of disease biology. We encourage the adoption of feed-forward networks based deep learning method in the metabolomics research community for classification.

ObjectiveTo identify dysregulated metabolic pathways in amyotrophic lateral sclerosis (ALS) versus control participants through untargeted metabolomics.MethodsUntargeted metabolomics was performed on plasma from ALS participants (n=125) around 6.8 months after diagnosis and healthy controls (n=71). Individual differential metabolites in ALS cases versus controls were assessed by Wilcoxon rank-sum tests, adjusted logistic regression and partial least squares-discriminant analysis (PLS-DA), while group lasso explored sub-pathway-level differences. Adjustment parameters included sex, age and body mass index (BMI). Metabolomics pathway enrichment analysis was performed on metabolites selected by the above methods. Finally, machine learning classification algorithms applied to group lasso-selected metabolites were evaluated for classifying case status.ResultsThere were no group differences in sex, age and BMI. Significant metabolites selected were 303 by Wilcoxon, 300 by logistic regression, 295 by PLS-DA and 259 by group lasso, corresponding to 11, 13, 12 and 22 enriched sub-pathways, respectively. ‘Benzoate metabolism’, ‘ceramides’, ‘creatine metabolism’, ‘fatty acid metabolism (acyl carnitine, polyunsaturated)’ and ‘hexosylceramides’ sub-pathways were enriched by all methods, and ‘sphingomyelins’ by all but Wilcoxon, indicating these pathways significantly associate with ALS. Finally, machine learning prediction of ALS cases using group lasso-selected metabolites achieved the best performance by regularised logistic regression with elastic net regularisation, with an area under the curve of 0.98 and specificity of 83%.ConclusionIn our analysis, ALS led to significant metabolic pathway alterations, which had correlations to known ALS pathomechanisms in the basic and clinical literature, and may represent important targets for future ALS therapeutics.

No abstract

Background: Recent studies have reveled the presence of a complex fungal community (mycobiome) in the oral cavity. However, the role of oral mycobiome in dental caries and its interaction with caries-associated bacteria is not yet clear. Methods: Whole-mouth supragingival plaque samples from 30 children (6-10 years old) with no caries, early caries, or advanced caries were sequenced for internal transcribed spacer 2 (ITS-2). The mycobiome profiles were correlated with previously published bacteriome counterparts. Interaction among selected fungal and bacterial species was assessed by co-culture or spent media experiments. Results: Fungal load was extremely low. Candida, Malassezia, Cryptococcus, and Trichoderma spp. were the most prevalent/abundant taxa. Advanced caries was associated with significantly higher fungal load and prevalence/abundance of Candida albicans. Cryptococcus neoformans and Candida sake were significantly over-abundant in early caries, while Malassezia globosa was significantly enriched in caries-free subjects. C. albicans correlated with Streptococcus mutans and Scardovia wiggsiae among other caries-associated bacteria, while M. globosa inversely correlated with caries-associated bacteria. In-vitro, M. globosa demonstrated inhibitory properties against S. mutans. Conclusions: the results substantiate the potential role of the oral mycobiome, primarily Candida species, in dental caries. Inter-kingdom correlations and inhibition of S. mutans by M. globosa are worth further investigation.

As of today (7 April 2020), more than 81,000 people around the world have died from the coronavirus disease 19 (COVID-19) pandemic. There is no approved drug or vaccine for COVID-19, although more than 10 clinical trials have been launched to test potential drugs. In an urgent response to this pandemic, I developed a bioinformatics pipeline to identify compounds and drug candidates to potentially treat COVID-19. This pipeline is based on publicly available single-cell RNA sequencing (scRNA-seq) data and the drug perturbation database “Library of Integrated Network-Based Cellular Signatures” (LINCS). I developed a ranking score system that prioritizes these drugs or small molecules. The four drugs with the highest total score are didanosine, benzyl-quinazolin-4-yl-amine, camptothecin, and RO-90-7501. In conclusion, I have demonstrated the utility of bioinformatics for identifying drugs than can be repurposed for potentially treating COVID-19 patients.

No abstract

Lilikoi (Hawaiian word for passion fruit) is a new and comprehensive R package for personalized pathway based classification modelling, using metabolomics data. Four basic modules are presented as the backbone of the package: 1) Feature mapping module, which standardizes the metabolite names provided by users, and map them to pathways. 2) Dimension transformation module, which transforms the metabolomic profiles to personalized pathway-based profiles using pathway deregulation scores (PDS). 3) Feature selection module which helps to select the significant pathway features related to the disease phenotypes, and 4) Classification and prediction module which offers various machine-learning classification algorithms. The package is freely available under the GPLv3 license through the github repository at: https://github.com/lanagarmire/lilikoi

Objective: The global rise in type 2 diabetes is associated with a concomitant increase in diabetic complications. Diabetic polyneuropathy is the most frequent type 2 diabetes complication and is associated with poor outcomes. The metabolic syndrome has emerged as a major risk factor for diabetic polyneuropathy; however, the metabolites associated with the metabolic syndrome that correlate with diabetic polyneuropathy are unknown. Methods: We conducted a global metabolomics analysis on plasma samples from a subcohort of participants from the Danish arm of Anglo-Danish-Dutch study of Intensive Treatment of Diabetes in Primary Care (ADDITION-Denmark) with and without diabetic polyneuropathy versus lean control participants. Results: Compared to lean controls, type 2 diabetes participants had significantly higher HbA1c (p = 0.0028), BMI (p = 0.0004), and waist circumference (p = 0.0001), but lower total cholesterol (p = 0.0001). Out of 991 total metabolites, we identified 15 plasma metabolites that differed in type 2 diabetes participants by diabetic polyneuropathy status, including metabolites belonging to energy, lipid, and xenobiotic pathways, among others. Additionally, these metabolites correlated with alterations in plasma lipid metabolites in type 2 diabetes participants based on neuropathy status. Further evaluating all plasma lipid metabolites identified a shift in abundance, chain length, and saturation of free fatty acids in type 2 diabetes participants. Importantly, the presence of diabetic polyneuropathy impacted the abundance of plasma complex lipids, including acylcarnitines and sphingolipids. Interpretation: Our explorative study suggests that diabetic polyneuropathy in type 2 diabetes is associated with novel alterations in plasma metabolites related to lipid metabolism.