0_PERMANOVA.Rmd

---
title: "PERMANOVA"
author: "NGG"
date: "07/09/2020"
output: html_document
---

# 1. Load libraries

```{r}
suppressWarnings(suppressMessages(library(tidyverse)))
suppressWarnings(suppressMessages(library(RColorBrewer)))
# suppressWarnings(suppressMessages(library(ggnewscale)))
# suppressWarnings(suppressMessages(library(ggthemes)))
# suppressWarnings(suppressMessages(library(ggrepel)))
# suppressWarnings(suppressMessages(library(ggpubr)))
suppressWarnings(suppressMessages(library(vegan)))

```

# 2. Import data

I am using filtered data from subsequent analyses. If you didn't filter your data, change file name accordingly.

```{r}
# Import data

# Feature_correspondence
features <- read.csv("../featureCorrespondence_MS2.csv", row.names = 1)
features <- as.data.frame(t(features))
# distance matrix - bray
dm <- vegdist(features, method="bray")

# Metadata
metadata <- read.csv("../metadata.csv")
metadata$SampleCode <- gsub("-", '.', metadata$SampleCode, fixed = T)

# MS2
ms2 <- read.csv("../../2_ms2_annotation/summary_ms2_annotation_energetics.csv")
# ms2 <- ms2 %>% 
#   select('Features', 'class', 'subclass', 'GFE')
```

# 3. PERMANOVA

Permutational Multivariate Analysis of Variance Using Distance Matrices
?adonis

Sources: https://rdrr.io/rforge/vegan/man/adonis.html, https://rpubs.com/collnell/manova

```{r}
set.seed(456)
permanova <- adonis(dm ~LeafLife + OakType + OakType:LeafLife, data=metadata, permutations=999, method="bray")
permanova
# print(as.data.frame(permanova$aov.tab))
filename <- paste0("permanova_results.csv")
write.csv(as.data.frame(permanova$aov.tab), filename)
```

Which taxa/feature contribute most to the community differences?

https://mibwurrepo.github.io/Microbial-bioinformatics-introductory-course-Material-2018/multivariate-comparisons-of-microbial-community-composition.html

```{r}
test <- "LL."
set.seed(456)
# LEAFLIFE; adonis tests homogeneity of dispersion among groups
p.LL <- adonis(features ~ LeafLife, data=metadata, permutations=999, method="bray")
p.LL 
## Interpretation:
# The R-square value is the important statistic for interpreting Adonis as it gives you the effect size. For example an R-squared of 0.44 means that 44% of the variation in distances is explained by the grouping being tested. The p value tells you whether or not this result was likely a result of chance. A p value of 0.05 means that there is a 5% chance that you detected a difference between groups (however large or small) when indeed there was none (the null hypothesis). 

############# extract coefficients for Feature representation
# to get coefficients supply the features df instead of the matrix dm
# matrix of coefficients of the linear model, with rows representing sources of variation and columns representing species; each column represents a fit of a species abundance to the linear model. These are what you get when you fit one species to your predictors.
p.LL.features <- coefficients(p.LL)['LeafLife1',]
p.LL.features <- p.LL.features[rev(order(abs(p.LL.features )))]
p.LL.features <- as.data.frame(p.LL.features) 
colnames(p.LL.features) <- paste0(test, 'f.contrib')
p.LL.features <- signif(p.LL.features, 2)
p.LL.features$Features <- rownames(p.LL.features)
filename <- paste0("features_contribution_LeafLife.csv")
write.csv(p.LL.features, filename)

## NEGATIVE == DRIVING DECIDUOUS
## POSTIVE == DRIVING BREVIDECIDUOUS

```