forked from zeyaxue/speeding-up-sci-heatmap-barplot
-
Notifications
You must be signed in to change notification settings - Fork 0
/
taxa-summary.rmd
251 lines (217 loc) · 9.44 KB
/
taxa-summary.rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
---
title: "Speeding-up-science-metatranscriptomics-taxa-summary"
author: "Zeya Xue"
date: "5/9/2019"
output: rmarkdown::github_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
knitr::opts_chunk$set(fig.path = "figs/")
```
# Speeding up science metatranscriptomics taxa summary
* Written by Zhengyao "Zeya" Xue, [ORCID](https://orcid.org/0000-0002-4930-8212)
* The data files and R scripts can be found in this [GitHub repo](https://github.com/zeyaxue/speeding-up-science-binder)
* [Launch Binder](https://mybinder.org/v2/gh/zeyaxue/speeding-up-science-binder/master?urlpath=rstudio)
* [html version](https://github.com/zeyaxue/speeding-up-science-binder/blob/master/taxa-summary.md)
* Thumbnail of expected Heatmap
```{r echo=FALSE, out.width='50%'}
knitr::include_graphics("https://raw.githubusercontent.com/zeyaxue/speeding-up-science-binder/master/figs/taxa_heat_thumb.png")
```
* Thumbnail of expected bar plot
```{r echo=FALSE, out.width='50%'}
knitr::include_graphics("https://raw.githubusercontent.com/zeyaxue/speeding-up-science-binder/master/figs/unnamed-chunk-8-1.png")
```
* P.S. The demonstration shown here is using data from [this paper](https://aem.asm.org/content/84/1/e02026-17.short)
## Introduction
The starting point of the workflow is
+ A count table, normalized or not. Looks like this:
```{r echo=FALSE}
TabTPM <- read.table(file.path("example_data/sample_TPM.tsv"),
header = TRUE, sep = "\t")
head(TabTPM)
```
+ A annotation or taxonomy table
```{r echo=FALSE}
Tabanno <- read.table(file.path("example_data/sample_annotation_classifications.tsv"),
header = TRUE, sep = "\t", na.strings = "<NA>")
head(Tabanno)
```
+ A sample metainfo table
```{r echo=FALSE, warning=FALSE}
samdf <- read.csv(file.path("example_data/Samdf.csv"))
head(samdf)
```
## Load packages and setting up
```{r message=FALSE, warning=FALSE}
library(phyloseq);packageVersion("phyloseq")
library(DESeq2);packageVersion("DESeq2")
library(ggplot2)
library(reshape2)
library(superheat)
library(plyr)
library(dplyr)
library(RColorBrewer)
```
## Import files to create phyloseq object
```{r}
# The otu table slot of phyloseq object
TabTPM <- read.table(file.path("example_data/sample_TPM.tsv"),
header = TRUE, sep = "\t")
row.names(TabTPM) <- TabTPM$CDS_ID
TabTPM <- TabTPM[,-1] # remove CDS_ID column. User defined.
TabTPM <- as.matrix.data.frame(TabTPM)
# The tax table slot of phyloseq object
Tabanno <- read.table(file.path("example_data/sample_annotation_classifications.tsv"),
header = TRUE, sep = "\t", na.strings = "<NA>")
rownames(Tabanno) <- Tabanno$CDS_ID
Tabanno <- Tabanno[,c(-1,-2)] # remove CDS_ID and KOID columns. User defined.
Tabanno <- as.matrix.data.frame(Tabanno)
# The sample data slot of phyloseq object
samdf <- read.csv(file.path("example_data/Samdf.csv"))
rownames(samdf) <- samdf$SampleID
ps <- phyloseq(otu_table(TabTPM, taxa_are_rows = TRUE),
tax_table(Tabanno), sample_data(samdf))
ps # 20000 taxa and 4 samples
```
## Optional taxonomy level clean up
```{r warning=FALSE}
# Define function to get the deepest taxa assignment level
RECps <- function(ps) {
TaxTab2 <- as.data.frame(ps@tax_table)
list.s = as.character(TaxTab2$Species)
list.g = as.character(TaxTab2$Genus)
list.f = as.character(TaxTab2$Family)
list.o = as.character(TaxTab2$Order)
list.c = as.character(TaxTab2$Class)
list.p = as.character(TaxTab2$Phylum)
list.k = as.character(TaxTab2$Kingdom)
list.d = as.character(TaxTab2$Domain)
list.REC = character(length(as.character(TaxTab2$Domain)))
for(i in 1:dim(TaxTab2)[1]){
S = which(TaxTab2$Species[i] == "" | is.na(TaxTab2$Species[i]))
G = which(TaxTab2$Genus[i] == "" | is.na(TaxTab2$Genus[i]))
Fa = which(TaxTab2$Family[i] == "" | is.na(TaxTab2$Family[i]))
O = which(TaxTab2$Order[i] == "" | is.na(TaxTab2$Order[i]))
C = which(TaxTab2$Class[i] == "" | is.na(TaxTab2$Class[i]))
P = which(TaxTab2$Phylum[i] == "" | is.na(TaxTab2$Phylum[i]))
K = which(TaxTab2$Kingdom[i] == "" | is.na(TaxTab2$Kingdom[i]))
D = which(TaxTab2$Domain[i] == "" | is.na(TaxTab2$Domain[i]))
if(length(S) == 0){
list.REC[i] <- list.s[i]
} else if(length(G) == 0){
list.REC[i] <- list.g[i]
} else if(length(Fa) == 0){
list.REC[i] <- list.f[i]
} else if(length(O) == 0){
list.REC[i] <- list.o[i]
} else if(length(C) == 0){
list.REC[i] <- list.c[i]
} else if(length(P) == 0){
list.REC[i] <- list.p[i]
} else if(length(K) == 0){
list.REC[i] <- list.k[i]
} else if(length(D) == 0){
list.REC[i] <- list.d[i]
} else {
list.REC[i] <- "meow"
}
}
TaxTab2$REC <- list.REC
TaxTab2$REC <- factor(TaxTab2$REC)
phyloseq(otu_table(ps), sample_data(ps),
TaxTab2 %>% as.matrix() %>% tax_table())
}
ps.REC <- RECps(ps)
ps.REC # 20000 taxa and 4 samples
```
## Heat map
```{r, fig.height = 13, fig.width = 10}
# Clean up the taxonomy
ps.REC.glom <- ps.REC %>% tax_glom(taxrank = "REC", NArm = FALSE)
# Run the next line if want relative abundance
ps.REC.per <- ps.REC.glom %>% transform_sample_counts(function(x) x/sum(x) )
taxa.df <- psmelt(ps.REC.per) # melt ps object
# aggregate for REC level plot
taxa.agg <- aggregate(Abundance ~ REC + SampleID,
data = taxa.df,
mean)
taxa.cast <- dcast(taxa.agg, REC ~ SampleID, mean, value.var = "Abundance")
# Define palette
my_palette <- colorRampPalette(c("red", "yellow", "green"))(n = 299)
# defines the color breaks manually for a "skewed" color transition
col_breaks = c(seq(-1,0,length=100), # for red
seq(0.01,0.8,length=100), # for yellow
seq(0.81,1,length=100)) # for green
# only plot the top 30 most abundant taxa
# need to change results from factor to numeric because of R
row.names(taxa.cast) <- taxa.cast$REC
taxa.cast <- taxa.cast[, -1]
indx <- sapply(taxa.cast, is.factor)
taxa.cast[indx] <- lapply(taxa.cast[indx], function(x) as.numeric(as.character(x)))
taxa.cast30 <- cbind(taxa.cast, total = rowSums(taxa.cast)) # need numeric values
taxa.cast30$taxa <- rownames(taxa.cast30)
taxa.cast30 <- head(arrange(taxa.cast30,desc(total)), n = 30)
row.names(taxa.cast30) <- taxa.cast30$taxa
taxa.cast30 <- taxa.cast30[, -c(5,6)] # remove total and taxa name colums
superheat(taxa.cast30,
# retain original order of rows/cols
pretty.order.rows = TRUE,
pretty.order.cols = TRUE,
row.dendrogram = TRUE,
col.dendrogram = TRUE,
grid.hline = TRUE,
row.title = "Annotation",
column.title = "SampleID",
left.label.text.size = 4,
bottom.label.text.size = 5,
left.label.size = 0.5,
# change the grid color to white (more pretty on a dark background)
grid.hline.col = "white",
grid.vline.col = "white")
```
## Stack bar plot
```{r}
# Clean up the taxonomy
ps.REC.glom <- ps.REC %>% tax_glom(taxrank = "REC", NArm = FALSE)
# Run the next line if want relative abundance
ps.REC.per <- ps.REC.glom %>% transform_sample_counts(function(x) x/sum(x) )
taxa.df <- psmelt(ps.REC.per) # melt ps object
# aggregate for REC level plot
taxa.agg <- aggregate(Abundance ~ REC + SampleID,
data = taxa.df,
mean)
# Get the names of the most abundant 15 taxa
ps.Notop15 <- prune_taxa(names(sort(taxa_sums(ps.REC.per), TRUE)[16:nrow(ps.REC.per@tax_table)]), ps.REC.per)
taxa_names_filt <- ps.Notop15@tax_table[,8] %>% as.character() # 8 for REC level
# convert REC colum to a character vector from a factor because R
taxa.agg$REC <- as.character(taxa.agg$REC)
# change the less abundant taxa names to "Other"
taxa.agg[taxa.agg$REC %in% taxa_names_filt,]$REC <- "Other"
# Set colors for plotting
mycol = colorRampPalette(brewer.pal(12, "Paired"))(16)
# Set levels of taxon for pretty plots
## I do not know this beforehand, modified after 1st generating plot to know the
## taxa names
taxa.agg$REC = factor(taxa.agg$REC, levels = c("Alteromonas_macleodii",
"Anaerophaga_thermohalophila",
"Aureispira_sp._CCB-QB1",
"Bacteroides_fragilis",
"Dyadobacter_alkalitolerans",
"Escherichia_coli",
"Haliscomenobacter_hydrossis",
"Lacinutrix_himadriensis",
"Lewinella_cohaerens",
"Nitrosomonas_communis",
"Phaeodactylibacter_xiamenensis",
"Salinibacter_ruber",
"Saprospira_grandis",
"Synechococcus_sp._BL107",
"Synechococcus_sp._CC9605",
"Other"))
ggplot(taxa.agg, aes(x = SampleID, y = Abundance, fill = REC)) +
geom_bar(stat = "identity") + #position = "fill" is for making the bar 100%
scale_fill_manual(values = mycol)+
theme(axis.title.x = element_blank()) + # Remove x axis title
guides(fill = guide_legend(reverse = FALSE, keywidth = 1, keyheight = 1)) +
ylab("Relative Abundance \n")
```