Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/hilldr/tsai_lgr5_endoderm_2015
Analysis scripts pertaining to the manuscript "LGR4 and LGR5 function redundantly during human endoderm differentiation" by Yu-Hwai Tsai et. al 2015
https://github.com/hilldr/tsai_lgr5_endoderm_2015
Last synced: about 1 month ago
JSON representation
Analysis scripts pertaining to the manuscript "LGR4 and LGR5 function redundantly during human endoderm differentiation" by Yu-Hwai Tsai et. al 2015
- Host: GitHub
- URL: https://github.com/hilldr/tsai_lgr5_endoderm_2015
- Owner: hilldr
- Created: 2015-11-18T20:35:40.000Z (about 9 years ago)
- Default Branch: master
- Last Pushed: 2015-11-20T15:55:00.000Z (about 9 years ago)
- Last Synced: 2023-02-28T08:41:11.861Z (almost 2 years ago)
- Language: R
- Size: 20.7 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
LGR4 and LGR5 function redundantly during human endoderm differentiation
========================================================================
Yu-Hwai Tsai1, David R. Hill1, Namit Kumar4, Sha Huang1, Briana
R. Dye2, Michael P. Verzi4 and Jason R. Spence1,2,3,*
1Division of Gastroenterology, Department of Internal Medicine
2Department of Cell and Developmental Biology
3Center for Organogenesis University of Michigan Medical School, Ann
Arbor, Michigan, USA, 48109
4Department of Genetics Rutgers, The State University of New Jersey,
Piscataway, New Jersey 08854
*Author for correspondence:
@TheSpenceLab
Author Contributions: YHT and JRS conceived the study YHT, DRH, SH, BRD,
NK and MPV conducted experiments All authors analyzed data JRS wrote the
manuscript All authors edited and approved final manuscript
Summary
=======
The Lgr family of transmembrane proteins (Lgr4, 5, 6) act as functional
receptors for R-spondin proteins (Rspo 1, 2, 3, 4), and have been shown
to potentiate Wnt signaling in different contexts. Lgr5, arguably the
best characterized of the Lgr family members, was originally identified
as a marker of crypt base columnar (CBC) stem cells of the adult murine
intestine, and has subsequently been show to act as a stem cell marker
in many contexts. Moreover, Lgr members play functionally redundant
roles in CBC maintenance. Here, we show that LGR5 is part of the human
definitive endoderm (DE) “gene signature”, as LGR5 transcripts are
robustly induced when human pluripotent stem cells (hPSCs) are
differentiated into DE. Our results demonstrate that LGR4 and 5 function
redundantly during human endoderm induction. Consistent with data in
human DE, we observe Lgr5 reporter (eGFP) activity in the E8.5 mouse
endoderm, and show the ability to linage trace Lgr5-creER cells labeled
as early as E8.5 into the adult intestine. ChIPseq shows that LGR5 is
directly bound by *β*-catenin early during endoderm induction, and
functional experiments support a role for LGR proteins as potentiators
of WNT signaling during DE induction. Taken together, our results show
that LGR proteins are induced in DE, are functionally required for DE
induction, and that they function to potentiate WNT signaling during
this process.
Bioinformatics methods and results
==================================
Required hardware and software
------------------------------
### Hardware
The analysis machine has the following characteristics:
- OS: Debian 7.9 wheezy
- Kernel: x86~64~ Linux 3.2.0-4-amd64
- Shell: bash 4.2.37
- CPU: Intel Xeon CPU E5504 @ 2GHz
- GPU: Quadro FX 380
- RAM: 48340MB
### Software
The code is intended for the UNIX environment. The following software is
required to execute the analysis
- FastQC v0.11.3
- Bowtie 2 version 2.2.5
- cufflinks v2.2.1
- R version 3.2.2 (2015-08-14) -- "Fire Safety"
Scripts and github page prepared using GNU Emacs 24.5.1 with Org-mode
version 8.2.10
#### R package dependencies
``` {.r .rundoc-block rundoc-language="R" rundoc-session="*R*" rundoc-eval="yes" rundoc-exports="code" rundoc-tangle="./src/expression_analysis.R"}
# R package dependencies -------------------------------------------------------
# Comment this line out if your machine already has these packages installed
install.packages(c("matrixStats","ggplot2","MASS","scales"),repos='https://watson.nci.nih.gov/cran_mirror/')
```
Tuxedo suite pipeline script
----------------------------
### Header
``` {.bash .rundoc-block rundoc-language="sh" rundoc-export="code" rundoc-eval="no" rundoc-tangle="./src/tuxedo_complete.sh"}
#! /bin/bash
GENES=/data/genomes/hg19_genes_refFlat.gtf
GENOME=/data/genomes/hg19.fa
INDEX=/data/genomes/hg19
```
### FastQC processing
``` {.bash .rundoc-block rundoc-language="sh" rundoc-export="code" rundoc-eval="no" rundoc-tangle="./src/tuxedo_complete.sh"}
#FASTQC quality control report generator
mkdir ../DATA/QC
for file in ../DATA/FASTQ/*.fastq* # will output filename as "$file"
do
FILENAME="$file"
fastqc --outdir=../DATA/QC $FILENAME
done
```
### Tophat/Bowtie processing
``` {.bash .rundoc-block rundoc-language="sh" rundoc-export="code" rundoc-eval="no" rundoc-tangle="./src/tuxedo_complete.sh"}
#! /bin/bash
mkdir ../DATA/BAM
for file in ../DATA/FASTQ/*.fastq*
do
SHORTNAME=$(basename "$file")
NAME2="${SHORTNAME##*/}"
DIRNAME="${NAME2%.*}"
tophat2 -p 8 --b2-very-sensitive --no-coverage-search --no-novel-juncs --GTF $GENES -o ../DATA/BAM/$DIRNAME $INDEX $file
done
```
### Cufflinks processing
``` {.bash .rundoc-block rundoc-language="sh" rundoc-export="code" rundoc-eval="no" rundoc-tangle="./src/tuxedo_complete.sh"}
#! /bin/bash
for d in ../DATA/BAM/*/
do
FILENAME="$file" #set variable FILENAME equal to file from line 1
SHORTNAME=$(basename "$file")
NAME2="${SHORTNAME##*/}"
DIRNAME="${d}"
cufflinks -p 8 -o $DIRNAME --multi-read-correct --compatible-hits-norm --upper-quartile-norm --GTF $GENES ${d}*hits.bam
done
```
#### Merged transcriptome setup file
``` {.bash .rundoc-block rundoc-language="sh" rundoc-export="code" rundoc-eval="no" rundoc-tangle="./src/gtf_assembly.txt"}
../DATA/BAM/Sample_ES1/transcripts.gtf
../DATA/BAM/Sample_ES2/transcripts.gtf
../DATA/BAM/Sample_ES3/transcripts.gtf
../DATA/BAM/Sample_DE1/transcripts.gtf
../DATA/BAM/Sample_DE2/transcripts.gtf
../DATA/BAM/Sample_DE3/transcripts.gtf
```
### Cuffmerge/Cuffquant processing
``` {.bash .rundoc-block rundoc-language="sh" rundoc-export="code" rundoc-eval="no" rundoc-tangle="./src/tuxedo_complete.sh"}
#! /bin/bash
cuffmerge -g $GENES -s $GENOME -p 8 -o ../DATA/merged_asm gtf_assembly.txt
for d in ../DATA/BAM/*/
do
FILENAME="$file" #set variable FILENAME equal to file from line 1
SHORTNAME=$(basename "$file")
NAME2="${SHORTNAME##*/}"
DIRNAME="${d}"
cuffquant -p 8 -o $DIRNAME --max-mle-iterations 100000 -v --multi-read-correct ../DATA/merged_asm/merged.gtf ${d}*hits.bam
done
```
### Cuffdiff/Cuffnorm processing
``` {.bash .rundoc-block rundoc-language="sh" rundoc-export="code" rundoc-eval="no" rundoc-tangle="./src/tuxedo_complete.sh"}
mkdir ../RESULTS
# CUFFNORM
cuffnorm -o ../RESULTS/normout -p 8 -L ES,DefEnd ../DATA/merged_asm/merged.gtf \
../DATA/BAM/Sample_ES1/abundances.cxb,../DATA/BAM/Sample_ES2/abundances.cxb,../DATA/BAM/Sample_ES3/abundances.cxb \
../DATA/BAM/Sample_DE1/abundances.cxb,../DATA/BAM/Sample_DE2/abundances.cxb,../DATA/BAM/Sample_DE3/abundances.cxb
```
R script
--------
### Import data from Cuffnorm output
``` {.r .rundoc-block rundoc-language="R" rundoc-session="*R*" rundoc-eval="yes" rundoc-exports="code" rundoc-tangle="./src/expression_analysis.R"}
## FPKM matrix input -----------------------------------------------------------
data1 <- read.table("../RESULTS/normout/genes.count_table",header=TRUE,sep="\t", stringsAsFactors = FALSE)
attr.table <- read.table("../RESULTS/normout/genes.attr_table",header=TRUE,sep="\t",stringsAsFactors = FALSE)
data1$gene_short_name <- attr.table$gene_short_name
# write out complete gene expression matrix
write.csv(data1, file="../RESULTS/ES&DE_dataset_cufflinksFPKM.csv")
```
### Generate row statistics
``` {.r .rundoc-block rundoc-language="R" rundoc-session="*R*" rundoc-eval="yes" rundoc-exports="code" rundoc-tangle="./src/expression_analysis.R"}
# Generate row stats --------------------------------------------------
# rename data1 as database to fork downstream additions
database <- data1
# generate group shortnames for selecting columns
ES <- grep("ES", colnames(database),ignore.case=F)
DE <- grep("DefEnd", colnames(database),ignore.case=F)
library(matrixStats)
## calculate mean by treatment type
database$es_mean <- rowMeans(database[,ES], na.rm=T)
database$de_mean <- rowMeans(database[,DE], na.rm=T)
# calculate log2 change
database$xlog2 <- log2(database$de_mean/database$es_mean)
## function to compare by row, returns t distribution
## The function is currently defined as
row.t <- function(mat1,mat2){
mat1 <- as.matrix(mat1)
mat2 <- as.matrix(mat2)
n1<-dim(mat1)[2]
n2<-dim(mat2)[2]
n<-n1+n2
m1<-rowMeans(mat1,na.rm=TRUE)
m2<-rowMeans(mat2,na.rm=TRUE)
v1<-rowVars(mat1,na.rm=TRUE)
v2<-rowVars(mat2,na.rm=TRUE)
vpool<-(n1-1)/(n-2)*v1 + (n2-1)/(n-2)*v2
tstat<-sqrt(n1*n2/n)*(m1-m2)/sqrt(vpool)
return(tstat)}
# calculate t-distribution for hES vs. Definitive endoderm
database$tstat <- row.t(database[,ES],database[,DE])
# express t-dist as p-value
database$p <- 2*pt(-abs(database$tstat),df=2)
# calculate Bonferroni correction
database$Bonf_p <- p.adjust(database$p, method = 'bonferroni', n = length(database$p))
database <- database[order(-database$xlog2),]
# output matrix with stats
write.csv(database,file="../RESULTS/ES&DE_dataset_cufflinksFPKM_stats.csv")
#--------------------------------------------------------------------------------
# SET CRITERIA FOR INCLUSION
# Fmpk cut-off
fco <- 0.4
# P-value cut-off
pco <- 0.05
#--------------------------------------------------------------------------------
#Count the number of reads with Fmpk > 2 in each group
database$es_count <- rowSums(database[,ES] >= fco)
database$de_count <- rowSums(database[,DE] >= fco)
# generate an output (present = 1, not present = 0) for each group
database$es_present <- ifelse(database$es_count > 1, 1, 0)
database$de_present <- ifelse(database$de_count > 1, 1, 0)
```
### "Volcano" expression plot
``` {.r .rundoc-block rundoc-language="R" rundoc-session="*R*" rundoc-eval="yes" rundoc-exports="code" rundoc-tangle="./src/expression_analysis.R"}
# Generate "Volcano" expression plot -------------------------------------------
# open png device
png(filename = "../RESULTS/volcano_plot.png",
width = 1200,
height = 1200,
units = "px",
pointsize =10,
bg = "white")
par(lwd=6,
mgp=c(9,3,0),
# c(bottom, left, top, right)
mar=c(12,15,1,1),
font.lab=2,
font.axis=1,
cex.lab=5,
font.main=2,
cex.axis=5,
bg="white")
# Axis Labels
ylab.name <- expression(paste("-log"[10],"(p-value)"))
main.lab <- ""
xlab.name <- expression(paste("Expression ratio DE/ES (log"[2]," FPKM)"))
# subset for plotting
# apply P-value cut-off
vplot <- subset(database, database$p <= pco)
# Gene expression must be present in at least one condition
vplot.1 <- subset(vplot[51:nrow(vplot),],
vplot$es_present ==1 | vplot$de_present == 1)
vplot.2 <- subset(database, database$p > pco)
# subset to genes expressed in both conditions
db <- subset(database, database$xlog2 != Inf|database$xlog2 != NA)
# subset top 20 up-regulated genes (formerly top 50)
db.50 <- db[1:20,]
# subset LGR5
lgr5 <- db[24,]
plot(vplot.1$xlog2,-log10(vplot.1$p),
type="n",
ylab=ylab.name,
xlab=xlab.name,
col=rgb(0,0,139,90,maxColorValue=255),
ylim= c(0,3.5),cex=2,xlim= c(-10,12))
grid(lwd=3)
lines(vplot.1$xlog2,-log10(vplot.1$p),
type="p",
col=rgb(65,105,225,100,maxColorValue=255),
ylim= c(0,5),
cex=2,
xlim= c(-10,12))
lines(vplot.2$xlog2,-log10(vplot.2$p),
type="p",
col=rgb(112,128,144,100,maxColorValue=255),
cex=2)
# points for top 20 upregulated
lines(db.50$xlog2,-log10(db.50$p),
type="p",
col=rgb(178,34,35,150,maxColorValue=255),
cex=2,
pch=21,
bg="white")
# pont for LGR5
lines(lgr5$xlog2,-log10(lgr5$p),
type="p",
col=rgb(178,34,35,150,maxColorValue=255),
cex=4,
pch=21,
bg="red")
n <- nrow(vplot)
# add arrow for LGR5
arrows(lgr5$xlog2-0.5,-log10(lgr5$p)+0.1,
x1 = lgr5$xlog2,y1 = -log10(lgr5$p),
lwd =5, col= "blue",length =0.1)
#Add gene name labels to top 30 upregulated genes
text(lgr5$xlog2-0.5,-log10(lgr5$p)+0.1,
labels = lgr5$gene_short_name,
col = "black",cex = 4,font = 4,pos = 2)
text(db.50$xlog2[1:20],(-log10(db.50$p)[1:20]),
labels = db.50$gene_short_name[1:20],
col="black",cex=2,font=2)
# Close plotting device
dev.off()
```
Top 50 up-regulated genes (Supplemental Table 1)
------------------------------------------------
``` {.r .rundoc-block rundoc-language="R" rundoc-session="*R*" rundoc-eval="yes" rundoc-exports="code" rundoc-tangle="./src/expression_analysis.R"}
# Supplemental Table 1 ---------------------------------------------------------
db <- subset(database, database$xlog2 != Inf|database$xlog2 != NA)
db.50 <- db[1:50,]
write.csv(db.50,file="../RESULTS/top50upregulated_ES&DE.csv")
```
Wnt gene expression plot and Supplemental Table 2
-------------------------------------------------
``` {.r .rundoc-block rundoc-language="R" rundoc-session="*R*" rundoc-eval="yes" rundoc-exports="code" rundoc-tangle="./src/expression_analysis.R"}
# Import Curated list of Wnt targets -------------------------------------------
list <- read.csv("../DATA/Wnt_Gene_List.csv",header=FALSE)
wnt.genes <- list[,1]
# retrieve rows with gene names matching wnt,genes
genes <- grep(paste(wnt.genes,collapse="|"),database$gene_short_name)
wnt.expression <- as.data.frame(database[genes,])
wnt.expression <- wnt.expression[order(wnt.expression$xlog2),]
write.csv(wnt.expression, file = "../RESULTS/wnt_gene_expression.csv") # Supplemental Table 2
wnt.expression <- subset(wnt.expression, wnt.expression$xlog2 != abs(Inf))
# sort by absolute value of log2 expression ratio (biggest differences first)
wnt.expression <- wnt.expression[order(-abs(wnt.expression$xlog2)),]
# Plot expression in ggplot2 ---------------------------------------------------
library(ggplot2)
library(MASS)
library(scales)
library(grid)
png(filename = "../RESULTS/wnt_scatter.png",
width = 1200,
height = 1200,
units = "px",
pointsize =10,
bg = "white")
print(ggplot(data = wnt.expression) +
geom_abline(intercept = 0 , slope = 1,
color = "grey", size = 2, linetype = "dashed") +
geom_point(shape = 21,alpha = I(0.95),
aes(de_mean, es_mean, size = -log(p,10),fill = xlog2)) +
guides(fill = guide_colorbar(title = expression(paste("DE/ES (log"[2]," FPKM)"))),
size = guide_legend(title = expression(paste("-log"[10],"(p-value)")))) +
scale_fill_gradient2(low = "blue", high = "red") +
scale_x_log10(breaks = trans_breaks("log10", function(x) 10^x),
labels = trans_format("log10", math_format(10^.x))) +
scale_y_log10(breaks = trans_breaks("log10", function(x) 10^x),
labels = trans_format("log10", math_format(10^.x))) +
annotation_logticks(size = 3) +
geom_text(data = wnt.expression[c(1:(grep("RSPO2",wnt.expression$gene_short_name)-1),(grep("RSPO2",wnt.expression$gene_short_name)+2):23),],
aes(de_mean,es_mean,label=gene_short_name),
size = 10, fontface = 2,
# nudge labels to improve visibility
vjust = ifelse(wnt.expression[c(1:(grep("RSPO2",wnt.expression$gene_short_name)-1),(grep("RSPO2",wnt.expression$gene_short_name)+2):23),]$gene_short_name == "TP53TG5" | wnt.expression[c(1:(grep("RSPO2",wnt.expression$gene_short_name)-1),(grep("RSPO2",wnt.expression$gene_short_name)+2):23),]$gene_short_name == "WIF1", -0.45, 0.5),
# move labels for 3 genes with 0 expression in DE out of view rather than plot names underneath y-axis (unreadable). Expression data available in Supplemental Table 2
hjust = ifelse(wnt.expression[c(1:(grep("RSPO2",wnt.expression$gene_short_name)-1),(grep("RSPO2",wnt.expression$gene_short_name)+2):23),]$de_mean < 1, 2, 0.5)) +
# Offset RSPO2 to improve visibility
geom_text(data = wnt.expression[grep("RSPO2",wnt.expression$gene_short_name),],
aes(de_mean,es_mean+7,label = gene_short_name),
size = 10,fontface=2) +
# Offset RSPO1 to improve visibility
geom_text(data = wnt.expression[grep("RSPO1",wnt.expression$gene_short_name),],
aes(de_mean+0.1,es_mean+2,label = gene_short_name),
size = 10,fontface=2) +
scale_size(range=c(0.25,25)) +
labs(x = "\nDefinitive Endoderm (Mean FPKM)",
y = "ES cells (Mean FPKM)\n") +
coord_fixed(ratio = 1) +
theme(panel.grid.minor = element_blank(),
panel.grid.major = element_blank(),
legend.position = "bottom",
legend.key.size = unit(2, "cm"),
legend.text = element_text(size = 20),
legend.title = element_text(size = 20, face =2),
axis.text = element_text(size=30, color = "black", face = 2),
axis.title = element_text(size = 35, face = 2),
panel.background = element_rect(fill = "grey85", color = "black"))
) # close print window
dev.off()
```
Complete workflow (Sequence alignment --> Expression plots)
==============================================================
#! /bin/bash
# First execute tuxedo pipeline to generate FPKM data (this will take a long time)
./tuxedo_complete.sh
# Then execute the R script
Rscript expression_analysis.R