html page update

CoAbundantTaxa.Rmd

@@ -0,0 +1,85 @@
+---
+title: "Identify Co-abundant taxa groups"
+author: "Leo Lahti, Sudarshan Shetty et al."
+bibliography: 
+- bibliography.bib
+output:
+  BiocStyle::html_document:
+    number_sections: no
+    toc: yes
+    toc_depth: 4
+    toc_float: true
+    self_contained: true
+    thumbnails: true
+    lightbox: true
+    gallery: true
+    use_bookdown: false
+    highlight: haddock
+---
+<!--
+  %\VignetteIndexEntry{microbiome tutorial}
+  %\VignetteEngine{knitr::rmarkdown}
+  %\usepackage[utf8]{inputenc}
+  %\VignetteEncoding{UTF-8}
+-->
+
+**Co-abundant groups of taxa**    
+
+Get data  
+```{r cag-1}
+library(microbiome)
+pseq <- atlas1006
+pseq.rel <- microbiome::transform(pseq, "compositional")
+```
+
+Identify co-abundant taxa. 
+```{r cag-2}
+# get median abundances 
+median_abund = apply(abundances(pseq.rel), MARGIN = 1, FUN = median)
+
+# select taxa with median abundance of more than or equal to 0.01
+abundant_taxa = abundances(pseq)[median_abund >= 0.01, ]
+# check how many taxa
+nrow(abundant_taxa)
+# check which taxa
+rownames(abundant_taxa)
+```
+
+
+Use Spearman's correlation to identify which of the co-abundant groups show correlation.   
+```{r cag-1b}
+# cor() is from `stats` package
+spearman_matrix = cor(t(abundant_taxa), method = "spearman")
+# hclust() is from `stats` package
+spearman_tree = hclust(dist(spearman_matrix), method = "ward")
+plot(spearman_tree)
+
+```
+
+Using the `heat` function in `microbiome pkg` now compare the correlation with the tree above.  
+```{r plot-heat}
+spearman_long_df <- reshape2::melt(spearman_matrix)
+head(spearman_long_df)
+
+heat(spearman_long_df, "Var1","Var2", 
+     order.rows = TRUE, order.cols = TRUE) + 
+  theme_bw() + theme(axis.text.x = element_text(angle = 90))
+```
+
+Validation of CAGs   
+```{r}
+# Validation of CAGs 
+# Randomly split the dataset, compute a correlation matrix and run a Mantel test
+subsample = sample(c(1:300), size = 220, replace = F)
+train_data = abundant_taxa[,subsample]
+test_data = abundant_taxa[,-subsample]
+train_spearman = cor(t(train_data), method = "spearman")
+test_spearman = cor(t(test_data), method = "spearman")
+vegan::mantel(as.dist(train_spearman), as.dist(test_spearman), permutations=999)
+```
+
+Further reading:  
+
+[de la Cuesta-Zuluaga, J., Corrales-Agudelo, V., Velásquez-Mejía, E.P. et al. Gut microbiota is associated with obesity and cardiometabolic disease in a population in the midst of Westernization. Sci Rep 8, 11356 (2018).](https://doi.org/10.1038/s41598-018-29687-x)   
+
+[Claesson, M., Jeffery, I., Conde, S. et al. Gut microbiota composition correlates with diet and health in the elderly. Nature 488, 178–184 (2012).](https://doi.org/10.1038/nature11319)