Paper reading (七十三)：A comparative study of gut microbiota in immune-mediated inflammatory diseases

论文题目：A comparative study of the gut microbiota in immune-mediated inflammatory diseases—does a common dysbiosis exist?

scholar 引用：12

页数：15

发表时间：2018.12

发表刊物：Microbiome 

作者：Jessica D. Forbes, Chih-yu Chen, ..., Gary Van Domselaar 

摘要：

Background
Immune-mediated inflammatory disease (IMID) represents a substantial health concern. It is widely recognized that IMID patients are at a higher risk for developing secondary inflammation-related conditions. While an ambiguous etiology is common to all IMIDs, in recent years, considerable knowledge has emerged regarding the plausible role of the gut microbiome in IMIDs. This study used 16S rRNA gene amplicon sequencing to compare the gut microbiota of patients with Crohn’s disease克罗恩病 (CD; N = 20), ulcerative colitis溃疡性结肠炎 (UC; N = 19), multiple sclerosis多发性硬化症 (MS; N = 19), and rheumatoid arthritis类风湿性关节炎 (RA; N = 21) versus healthy controls健康对照组 (HC; N = 23). Biological replicates were collected from participants within a 2-month interval. This study aimed to identify common (or unique) taxonomic biomarkers of IMIDs using both differential abundance testing and a machine learning approach.
Results
Significant microbial community differences between cohorts were observed (pseudo F = 4.56; p = 0.01). Richness and diversity were significantly different between cohorts (pFDR < 0.001) and were lowest in CD while highest in HC. Abundances of Actinomyces, Eggerthella, Clostridium III, Faecalicoccus, and Streptococcus (pFDR < 0.001) were significantly higher in all disease cohorts relative to HC, whereas significantly lower abundances were observed for Gemmiger, Lachnospira, and Sporobacter (pFDR < 0.001). Several taxa were found to be differentially abundant in IMIDs versus HC including significantly higher abundances of Intestinibacter in CD, Bifidobacterium in UC, and unclassified Erysipelotrichaceae in MS and significantly lower abundances of Coprococcus in CD, Dialister in MS, and Roseburia in RA. A machine learning approach to classify disease versus HC was highest for CD (AUC = 0.93 and AUC = 0.95 for OTU and genus features, respectively) followed by MS, RA, and UC. Gemmiger and Faecalicoccus were identified as important features for classification of subjects to CD and HC. In general, features identified by differential abundance testing were consistent with machine learning feature importance.
Conclusions
This study identified several gut microbial taxa with differential abundance patterns common to IMIDs. We also found differentially abundant taxa between IMIDs. These taxa may serve as biomarkers for the detection and diagnosis of IMIDs and suggest there may be a common component to IMID etiology.

正文组织架构：

1. Background

2. Methods

2.1 Patients and stool sample collection

2.2 DNA extraction, library preparation, and 16S rRNA gene amplicon sequencing

2.3 Data analysis

2.4 Evaluation of Gram-negative abundance

2.5 Processing and statistical analysis of operational taxonomic units

2.6 Machine learning classification

2.7 Diversity

3. Results

3.1 Community structure, richness, and diversity of the gut microbiota

3.2 Taxonomic composition of the gut microbiota

3.3 Model performance of machine learning classifiers

3.4 Feature importance of the classifiers

4.  Discussion

5. Conclusions

正文部分内容摘录：

1. Biological Problem: What biological problems have been solved in this paper?

• This study aimed to identify common (or unique) taxonomic biomarkers of IMIDs using both differential abundance testing and a machine learning approach.
• This study identified several gut microbial taxa with differential abundance patterns common to IMIDs.
•  

2. Main discoveries: What is the main discoveries in this paper?

• introduction：Differential abundance testing and a supervised machine learning approach using a random forest algorithm were conducted and results compared to identify taxonomic biomarkers common among or specific to IMIDs.
• abstract:
• Significant microbial community differences between cohorts were observed (pseudo F = 4.56; p = 0.01).
• A machine learning approach to classify disease versus HC was highest for CD (AUC = 0.93 and AUC = 0.95 for OTU and genus features, respectively) followed by MS, RA, and UC.
• Gemmiger and Faecalicoccus were identified as important features for classification of subjects to CD and HC.
• In general, features identified by differential abundance testing were consistent with machine learning feature importance.
• We also found differentially abundant taxa between IMIDs.
• conclusion:
• We conclude that the composition of the gut microbiota is altered in CD, UC, MS, and RA, especially relating to varying degrees of gut dysbiosis that were evident. 
• we also show that within IMIDs, several microorganisms demonstrated significant compositional differences. 
• We have uncovered several microorganisms that are consistently higher (or lower) in IMID relative to HC using differential abundance testing and machine learning, suggesting that there may be common microbial taxonomic biomarkers for IMID. 
• discussion:  超长。。。13段的discussion就问你们有没有见过。。。很多关于观察结合的偏生物的分析
• The results of this study are twofold.
• First, we found important differences in the stool microbial profile in IMID relative to health using differential abundance analyses and machine learning using the random forest algorithm.
• Second, we showed how microbial populations vary between different IMIDs.

3. ML(Machine Learning) Methods: What are the ML methods applied in this paper?

• a total of 7943 OTUs, corresponding to 426 genera, was obtained
• In our random forest classifiers for CD compared to HC, a Roseburia OTU was identified as an important feature (higher abundance in HC), suggesting its importance for health. 
• we conducted machine learning using a random forest approach to conduct binary classification of diseases versus HC using OTU or genus abundance as features. 
• Random forest machine learning classification was conducted using the “randomForest” R package with 500 trees and the number of randomly sampled variables as the square root of the variable counts. Classifiers were built based on samples from the first time point and excluded technical replicates to avoid inflation of performance evaluation.
• We evaluated both OTU and genus OTU abundances as training data features to assess the performance at different levels of taxonomic specificity. 

4. ML Advantages: Why are these ML methods better than the traditional methods in these biological problems?

• In our study, we chose to evaluate the feature importance of OTU and genus classifiers, since they can in theory differ in their ability to classify disease state versus HC. 
• our CD versus HC classifiers performed slightly better than previously reported classifiers for IBD (CD and UC combined) versus HC
• In contrast, our multi-disease classifier performed poorly, which is expected considering similar poor performances were observed for our binary classifiers between different diseases, such as MS versus RA, CD versus RA, and CD versus UC. 
• These results suggest that the similarities in gut microbiota composition between diseases make their classification difficult. 

5. Biological Significance: What is the biological significance of these ML methods’ results?

• As each tree in the classifier is fit from a bootstrap sample set from the training data, samples that were not used to fit the corresponding tree were used to calculate the out-of-bag (OOB) error to assess model performance. For each pairwise comparison, the average OOB performance from ten models with different seeds was reported as balanced accuracy (BA) and area under the receiver operating characteristic curve (AUC).
• Feature importance averaged from the ten runs was reported as a measure of the mean decrease in Gini index (the impurity function) and visualized using the “ComplexHeatmap” R package

6. Prospect: What are the potential applications of these machine learning methods in biological science?

• These taxa may serve as biomarkers for the detection and diagnosis of IMIDs and suggest there may be a common component to IMID etiology.
• Further research into these microorganisms and their associated functions within the host are needed in order to establish any causality in disease pathogenesis and future therapeutic potential.
• These study findings reveal that microbiota characterization can be performed using “imperfect” datasets, which may be further overcome through the use of robust machine learning approaches.
• Other limitations of this study include the use of stool rather than mucosa (which houses distinct and more immunologically related microorganisms) and the targeted evaluation of the bacterial component of the microbiome rather than including microorganisms such as fungi or viruses.
• Importantly, our study was also limited by the lack of certain patient-related data.

7. Mine Question(Optional)