PHAR350 – Bioinformatics II

PHAR350 – Bioinformatics II
MetaboAnalyst: Assessment

Dataset description: Investigation of chronic kidney disease using metabolomics

Chronic kidney disease (CKD) leads to a decreased sensitivity to the metabolic effects of insulin. In this study, researchers examined the blood plasma metabolome in 95 adults without diabetes in the fasted state (58 with moderate-severe CKD, 37 with normal glomerular filtration rate), of whom 60 had plasma samples collected during a hyperinsulinemic-euglycemic clamp (40 with CKD). They assessed heterogeneity in the response to insulin in order to identify potential cellular metabolic pathways linking CKD with insulin resistance. Class: 0 = control patient, 1= CKD patient.

Task: Please analyse the dataset that you received via email by answering to the three parts, each of them described below and corresponding to a specific analysis in MetaboAnalyst. Be as specific as possible, and don’t forget to respond with inserted Figures whenever requested.

First Part: Statistical analysis

1. Download the dataset, and chose « spectral bins » and « sample in row (unpaired) ». Submit and pass the data filtering section: choose “none” in filtering.
2. Normalisation: Find the best normalisation method. Which sample normalization and data scaling did you use? Add the Figure showing the normalized result for both feature and sample normalizations.
3. PCA analysis: do you see, by examining the 2D score-plot, any clear outliers? If yes, put the 2D PCA figure here to show it, remove the outlier(s) and perform a new normalisation on the dataset without the outlier(s). Again, which sample normalization and data scaling did you use? Add the Figure showing the normalized result for both feature and sample normalization to your report.
4. PCA analysis: Show the 2D scores plot.
5. PLS-DA analysis: Show the 2D scores plot, and record the R² and Q² values: Do you think these are satisfactory values? Please explain.
6. Name the more important variables (from ‘imp. Features’; i.e. those with a VIP value > 1.2).
   Which significant variables are higher for the CKD patient (“1”), and which of these are higher for the control patients (“0”)?
7. Show the permutation test: What is the p-value obtained? What does this mean?

Second Part: Enrichment analysis

1. Perform an enrichment analysis using the over representation analysis. Type the name of the metabolites (one metabolite per line) that you identified as significant in step 5 of the previous question. Click submit, and please make sure that each metabolite has a correct hit during the “Compound Name/ID Standardization”. Click submit again.
2. I know that some of you don’t have a human dataset, but one with animals/plants, and so you shouldn’t be using the enrichment analysis since it’s supposed to be usable only with human data. However, in this case, do it anyway so you will have done it yourself with your own dataset, and click on “Pathway-associated metabolite sets (SMPDB)”.

• Insert the image of the metabolite set enrichment overview (use the bar chart view, not the network view).
• What are the most significant pathways, and what are the p values of these (note here only the pathways that are significant, based on their p values).

Third part: Pathway analysis

1. Perform a pathway analysis using the “one-column compound list” panel. Type the name of the metabolites (one metabolite per line) that you identified as significant in step 5 of the statistical analysis question (first part). Use “Compound Name” as Input Type. Click submit, and ensure that each metabolite has the correct hit during the “Compound Name/ID Standardization”. Click submit again.
2. Keep the “Hypergeometric Test” and “Relative-Betweeness Centrality” as the pathway algorithms. Select the correct organism (human/pig/plant…), and then click submit.

– Insert a Figure for each significant (only the significant ones, not the others) pathway that you have identified as significant based on their p values.

– What is the p value for these pathways?

– How is MetaboAnalyst determining that a pathway is significant or not?

– What determines the size of the circle for each pathway?

– Take a closer look at each of your significant pathways; in these particular pathways, which of your metabolites is at a “bottleneck” or at a “hub” position?