Compare CENTIPEDE predictions for CTCF for regions that differentially open between hypoxia vs. normoxia

Last updated: 2018-07-25

workflowr checks: (Click a bullet for more information)

✔ R Markdown file: up-to-date

Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.
✔ Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.
✔ Seed: set.seed(20180613)

The command set.seed(20180613) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.
✔ Session information: recorded

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.
✔ Repository version: cbd5c0b
Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility. The version displayed above was the version of the Git repository at the time these results were generated.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:
```
Ignored files:
    Ignored:    .DS_Store
    Ignored:    .Rhistory
    Ignored:    .Rproj.user/
    Ignored:    code_RCC/.DS_Store
    Ignored:    data/.DS_Store

Untracked files:
    Untracked:  analysis/compare_centipede_predictions_DiffACregions_CTCF.Rmd
    Untracked:  data/motifs/
    Untracked:  data/wgEncodeMapability/
    Untracked:  docs/figure/compare_centipede_predictions.Rmd/
    Untracked:  docs/figure/compare_centipede_predictions_DiffACregions_CTCF.Rmd/
    Untracked:  workflow_setup.R

Unstaged changes:
    Modified:   analysis/ATACseq_preprocessing_pipeline.Rmd
    Modified:   analysis/compare_centipede_predictions_HIF1A.Rmd
    Modified:   analysis/index.Rmd
    Modified:   code_RCC/fitCentipede_ATAC-seq_counts.R
```
Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.

Expand here to see past versions:

File	Version	Author	Date	Message
Rmd	cbd5c0b	kevinlkx	2018-07-25	compare centipede predictions for HIF1A in DiffAC regions

library(ggplot2)
library(grid)
library(gridExtra)
suppressPackageStartupMessages(library(GenomicRanges))
library(limma)


Attaching package: 'limma'

The following object is masked from 'package:BiocGenerics':

    plotMA

library(edgeR)
library(VennDiagram)

Loading required package: futile.logger

message <- futile.logger::flog.threshold(futile.logger::ERROR, name = "VennDiagramLogger")

## venn diagram
plot_venn_overlaps <- function(overlaps.m, title = "", col_fill = NULL, category.names = NULL){
  grid.newpage()
  overlaps_venn.l <- lapply(as.data.frame(overlaps.m), function(x) which(x == 1))
  if(is.null(col_fill)){
    col_fill <-  1:length(overlaps_venn.l)
  }
  if(is.null(category.names)){
    category.names <- names(x)
  }
  
  venn.plot <- venn.diagram( 
    x = overlaps_venn.l,
    category.names = category.names, 
    filename = NULL,
    fill = col_fill,
    alpha=rep(0.5, length(overlaps_venn.l)), 
    cex = 1.5, 
    cat.fontface=4, 
    main=title) 
  grid.draw(venn.plot)
}

parameters

tf_name <- "HIF1A"
pwm_name <- "HIF1A_MA1106.1_1e-4"

thresh_PostPr_bound <- 0.99
cat(pwm_name, "\n")

HIF1A_MA1106.1_1e-4

flank <- 100

load diff accessibility test results, comparing hypoxia vs. normoxia.

log fold change > 0 indicates differentially open in hypoxia.
log fold change < 0 indicates differentially open in normoxia.

diffAC_regions.df <- read.csv("~/Dropbox/research/ATAC_DNase/ATAC-seq_Olivia_Gray/results/DiffAC_regions/ordered_results_withcoords.csv")

cat(nrow(diffAC_regions.df), "regions in total \n")

138927 regions in total

diffAC_regions.df <- diffAC_regions.df[, c("chr", "Start", "End","GeneID", "baseMean", "Strand",  "log2FoldChange", "lfcSE", "stat", "pvalue", "padj")]

diffAC_sig_regions.df <- diffAC_regions.df[diffAC_regions.df$padj < 0.1, ]
cat(nrow(diffAC_sig_regions.df), "significant regions \n")

2390 significant regions

hist(diffAC_regions.df$log2FoldChange, xlab = "log2FoldChange", main = "Differentially open regions (FDR < 10%)")

diffAC_sigH_regions.df <- diffAC_sig_regions.df[diffAC_sig_regions.df$log2FoldChange > 0, ]
cat(nrow(diffAC_sigH_regions.df), "regions differentially open in hypoxia. \n")

201 regions differentially open in hypoxia.

diffAC_sigN_regions.df <- diffAC_sig_regions.df[diffAC_sig_regions.df$log2FoldChange < 0, ]
cat(nrow(diffAC_sigN_regions.df), "regions differentially open in normoxia. \n")

2189 regions differentially open in normoxia.

diffAC_sig_regions.gr <- makeGRangesFromDataFrame(diffAC_sig_regions.df, start.field = "Start", end.field = "End", keep.extra.columns = T)

diffAC_sigH_regions.gr <- makeGRangesFromDataFrame(diffAC_sigH_regions.df, start.field = "Start", end.field = "End", keep.extra.columns = T)

diffAC_sigN_regions.gr <- makeGRangesFromDataFrame(diffAC_sigN_regions.df, start.field = "Start", end.field = "End", keep.extra.columns = T)

load CENTIPEDE predictions

dir_predictions <- paste0("~/Dropbox/research/ATAC_DNase/ATAC-seq_Olivia_Gray/results/centipede_predictions/", pwm_name)

## condition: N
bam_namelist_N <- c("N1_nomito_rdup.bam", "N2_nomito_rdup.bam", "N3_nomito_rdup.bam")

site_predictions_N.l <- vector("list", 3)
names(site_predictions_N.l) <- bam_namelist_N

for(i in 1:length(bam_namelist_N)){
  bam_basename <- tools::file_path_sans_ext(basename(bam_namelist_N[[i]]))
  site_predictions_N.l[[i]] <- read.table(paste0(dir_predictions, "/", pwm_name, "_", bam_basename, "_predictions.txt"), header = T, stringsAsFactors = F)
}

CentPostPr_N.df <- data.frame(N1 = site_predictions_N.l[[1]]$CentPostPr, 
                              N2 = site_predictions_N.l[[2]]$CentPostPr, 
                              N3 = site_predictions_N.l[[3]]$CentPostPr)

CentLogRatios_N.df <- data.frame(N1 = site_predictions_N.l[[1]]$CentLogRatios, 
                                 N2 = site_predictions_N.l[[2]]$CentLogRatios, 
                                 N3 = site_predictions_N.l[[3]]$CentLogRatios)

## condition: H
bam_namelist_H <- c("H1_nomito_rdup.bam", "H2_nomito_rdup.bam", "H3_nomito_rdup.bam")

site_predictions_H.l <- vector("list", 3)
names(site_predictions_H.l) <- bam_namelist_H

for(i in 1:length(bam_namelist_H)){
  bam_basename <- tools::file_path_sans_ext(basename(bam_namelist_H[[i]]))
  site_predictions_H.l[[i]] <- read.table(paste0(dir_predictions, "/", pwm_name, "_", bam_basename, "_predictions.txt"), header = T, stringsAsFactors = F)
}

CentPostPr_H.df <- data.frame(H1 = site_predictions_H.l[[1]]$CentPostPr, 
                              H2 = site_predictions_H.l[[2]]$CentPostPr, 
                              H3 = site_predictions_H.l[[3]]$CentPostPr)

CentLogRatios_H.df <- data.frame(H1 = site_predictions_H.l[[1]]$CentLogRatios, 
                                 H2 = site_predictions_H.l[[2]]$CentLogRatios, 
                                 H3 = site_predictions_H.l[[3]]$CentLogRatios)

if(any(site_predictions_N.l[[1]]$name != site_predictions_H.l[[1]]$name)){
  stop("sites not match!")
}

sites.df <- site_predictions_N.l[[1]][,1:7]

## get motif coordinates
if(sites.df[1, "end"] - sites.df[1, "start"] > flank){
  sites.df$start <- sites.df$start + flank
  sites.df$end <- sites.df$end - flank
}


sites.gr <- makeGRangesFromDataFrame(sites.df, start.field = "start", end.field = "end", keep.extra.columns = F)


CentPostPr.df <- cbind(CentPostPr_N.df, CentPostPr_H.df)
CentLogRatios.df <- cbind(CentLogRatios_N.df, CentLogRatios_H.df)

sites_CentPostPr.df <- cbind(sites.df, CentPostPr_N.df, CentPostPr_H.df)
sites_CentLogRatios.df <- cbind(sites.df, CentLogRatios_N.df, CentLogRatios_H.df)

intersect CENTIPEDE sites with diffAC regions

overlaps_diffAC.df <- as.data.frame(findOverlaps(query = sites.gr, subject = diffAC_sig_regions.gr, type = "within", ignore.strand = T))
idx_sites_diffAC <- unique(overlaps_diffAC.df$queryHits)
cat(length(idx_sites_diffAC), "candidate motif sites differentially open in hypoxia or normoxia. \n")

101 candidate motif sites differentially open in hypoxia or normoxia.

overlaps_sigH.df <- as.data.frame(findOverlaps(query = sites.gr, subject = diffAC_sigH_regions.gr, type = "within", ignore.strand = T))
idx_sites_sigH <- unique(overlaps_sigH.df$queryHits)

cat(length(idx_sites_sigH), "candidate motif sites differentially open in hypoxia. \n")

22 candidate motif sites differentially open in hypoxia.

overlaps_sigN.df <- as.data.frame(findOverlaps(query = sites.gr, subject = diffAC_sigN_regions.gr, type = "within", ignore.strand = T))
idx_sites_sigN <- unique(overlaps_sigN.df$queryHits)

cat(length(idx_sites_sigN), "candidate motif sites differentially open in normoxia. \n")

79 candidate motif sites differentially open in normoxia.

binarize to bound and unbound

cat("Number of bound sites that are differentially open in hypoxia: \n")

Number of bound sites that are differentially open in hypoxia:

colSums(CentPostPr.df[idx_sites_sigH, ] > thresh_PostPr_bound)

N1 N2 N3 H1 H2 H3 
14 12  9 21 18 22

cat("Number of bound sites that are differentially open in normoxia: \n")

Number of bound sites that are differentially open in normoxia:

colSums(CentPostPr.df[idx_sites_sigN, ] > thresh_PostPr_bound)

N1 N2 N3 H1 H2 H3 
79 79 78 45 36 53

Average binding probablity and average logRatios

all motif sites

# binding probablity
par(pty="s")
plot(rowMeans(CentPostPr_N.df), rowMeans(CentPostPr_H.df), 
     xlab = "N average P(Bound)", ylab = "H average P(Bound)", main = tf_name,
     pch = ".", col = rgb(0,0,1,0.7))
abline(a=0, b=1, col = "darkgray")

# logRatios
par(mfrow = c(1,2))
par(pty="s")
plot(rowMeans(CentLogRatios_N.df), rowMeans(CentLogRatios_H.df), 
     xlab = "N average logRatios", ylab = "H average logRatios", main = tf_name, 
     pch = ".", col = rgb(0,0,1,0.7))
abline(a=0,b=1,col = "darkgray")

plot(x = (rowMeans(CentLogRatios_H.df)+rowMeans(CentLogRatios_N.df))/2, 
     y = rowMeans(CentLogRatios_H.df) - rowMeans(CentLogRatios_N.df),
     xlab = "average logRatios", ylab = "Difference in logRatios (H - N)", main = tf_name,
     pch = ".", col = rgb(0,0,1,0.7))
abline(v=0, h=0, col = "darkgray")

sites that are differentially open in hypoxia

cat(length(idx_sites_sigH), "candidate motif sites differentially open in hypoxia. \n")

22 candidate motif sites differentially open in hypoxia.

# binding probablity
par(pty="s")
plot(rowMeans(CentPostPr_N.df[idx_sites_sigH,]), rowMeans(CentPostPr_H.df[idx_sites_sigH,]), 
     xlab = "N average P(Bound)", ylab = "H average P(Bound)", main = paste(tf_name, "bound sites"),
     pch = 20, col = rgb(0,0,1,0.7))
abline(a=0, b=1, col = "darkgray")

# logRatios
par(mfrow = c(1,2))
par(pty="s")
plot(rowMeans(CentLogRatios_N.df[idx_sites_sigH,]), rowMeans(CentLogRatios_H.df[idx_sites_sigH,]), 
     xlab = "N average logRatios", ylab = "H average logRatios", main = tf_name, 
     pch = 20, col = rgb(0,0,1,0.7))
abline(a=0,b=1,col = "darkgray")

plot(x = (rowMeans(CentLogRatios_H.df[idx_sites_sigH,])+rowMeans(CentLogRatios_N.df[idx_sites_sigH,]))/2, 
     y = rowMeans(CentLogRatios_H.df[idx_sites_sigH,]) - rowMeans(CentLogRatios_N.df[idx_sites_sigH,]),
     xlab = "average logRatios", ylab = "Difference in logRatios (H - N)", main = tf_name,
     pch = 20, col = rgb(0,0,1,0.7))
abline(v=0, h=0, col = "darkgray")

sites that are differentially open in normoxia

cat(length(idx_sites_sigN), "candidate motif sites differentially open in normoxia \n")

79 candidate motif sites differentially open in normoxia

# binding probablity
par(pty="s")
plot(rowMeans(CentPostPr_N.df[idx_sites_sigN,]), rowMeans(CentPostPr_H.df[idx_sites_sigN,]), 
     xlab = "N average P(Bound)", ylab = "H average P(Bound)", main = paste(tf_name, "bound sites"),
     pch = 20, col = rgb(0,0,1,0.7))
abline(a=0, b=1, col = "darkgray")

# logRatios
par(mfrow = c(1,2))
par(pty="s")
plot(rowMeans(CentLogRatios_N.df[idx_sites_sigN,]), rowMeans(CentLogRatios_H.df[idx_sites_sigN,]), 
     xlab = "N average logRatios", ylab = "H average logRatios", main = tf_name, 
     pch = 20, col = rgb(0,0,1,0.7))
abline(a=0,b=1,col = "darkgray")

plot(x = (rowMeans(CentLogRatios_H.df[idx_sites_sigN,])+rowMeans(CentLogRatios_N.df[idx_sites_sigN,]))/2, 
     y = rowMeans(CentLogRatios_H.df[idx_sites_sigN,]) - rowMeans(CentLogRatios_N.df[idx_sites_sigN,]),
     xlab = "average logRatios", ylab = "Difference in logRatios (H - N)", main = tf_name,
     pch = 20, col = rgb(0,0,1,0.7))
abline(v=0, h=0, col = "darkgray")

Compare logRatios for differentially accessible sites using limma

targets <- data.frame(bam = c(bam_namelist_N, bam_namelist_H), 
                      label = colnames(CentLogRatios.df), 
                      condition = rep(c("N", "H"), each = 3))

print(targets)

                 bam label condition
1 N1_nomito_rdup.bam    N1         N
2 N2_nomito_rdup.bam    N2         N
3 N3_nomito_rdup.bam    N3         N
4 H1_nomito_rdup.bam    H1         H
5 H2_nomito_rdup.bam    H2         H
6 H3_nomito_rdup.bam    H3         H

condition <- factor(targets$condition, levels = c("N", "H"))
design <- model.matrix(~0+condition)
colnames(design) <- levels(condition)
print(design)

  N H
1 1 0
2 1 0
3 1 0
4 0 1
5 0 1
6 0 1
attr(,"assign")
[1] 1 1
attr(,"contrasts")
attr(,"contrasts")$condition
[1] "contr.treatment"

CentLogRatios_diffAC.df <- CentLogRatios.df[idx_sites_diffAC, ]

fit <- lmFit(CentLogRatios_diffAC.df, design)
contrasts <- makeContrasts(H-N, levels=design)
fit2 <- contrasts.fit(fit, contrasts)
fit2 <- eBayes(fit2, trend=TRUE)
num_diffbind <- summary(decideTests(fit2))

percent_diffbind <- round(num_diffbind / sum(num_diffbind) * 100, 2)

cat(num_diffbind[1], "sites differentially open in normoxia (", percent_diffbind[1], "%) \n", 
    num_diffbind[3], "sites differentially open in hypoxia (", percent_diffbind[3], "%) \n",
    num_diffbind[2], "sites not significantly different (", percent_diffbind[2], "%) \n")

79 sites differentially open in normoxia ( 78.22 %) 
 14 sites differentially open in hypoxia ( 13.86 %) 
 8 sites not significantly different ( 7.92 %)

# volcanoplot(fit2, main="H vs. N", xlab = "Difference in logRatios (H - N)")

plot(x = fit2$coef, y = -log10(fit2$p.value),
     xlab = "Difference in logRatios (H - N)", ylab = "-log10(P-value)", main= paste(tf_name, "H vs. N"),
     pch = 16, cex = 0.35)

Session information

sessionInfo()

R version 3.3.3 (2017-03-06)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
Running under: macOS  10.13.4

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
 [1] stats4    parallel  grid      stats     graphics  grDevices utils    
 [8] datasets  methods   base     

other attached packages:
 [1] VennDiagram_1.6.17   futile.logger_1.4.3  edgeR_3.14.0        
 [4] limma_3.28.21        GenomicRanges_1.24.3 GenomeInfoDb_1.8.7  
 [7] IRanges_2.6.1        S4Vectors_0.10.3     BiocGenerics_0.18.0 
[10] gridExtra_2.3        ggplot2_2.2.1       

loaded via a namespace (and not attached):
 [1] Rcpp_0.12.14         pillar_1.1.0         git2r_0.21.0        
 [4] plyr_1.8.4           workflowr_1.0.1      XVector_0.12.1      
 [7] futile.options_1.0.0 R.methodsS3_1.7.1    R.utils_2.6.0       
[10] tools_3.3.3          zlibbioc_1.18.0      digest_0.6.13       
[13] evaluate_0.10.1      tibble_1.4.1         gtable_0.2.0        
[16] rlang_0.1.6          yaml_2.1.16          stringr_1.2.0       
[19] knitr_1.18           rprojroot_1.3-2      rmarkdown_1.9       
[22] lambda.r_1.1.9       magrittr_1.5         whisker_0.3-2       
[25] splines_3.3.3        backports_1.1.2      scales_0.4.1        
[28] htmltools_0.3.6      colorspace_1.3-2     stringi_1.1.5       
[31] lazyeval_0.2.1       munsell_0.4.3        R.oo_1.21.0

This reproducible R Markdown analysis was created with workflowr 1.0.1

Compare CENTIPEDE predictions for CTCF for regions that differentially open between hypoxia vs. normoxia

Kaixuan Luo

7/25/2018

parameters

load diff accessibility test results, comparing hypoxia vs. normoxia.

load CENTIPEDE predictions

intersect CENTIPEDE sites with diffAC regions

binarize to bound and unbound

Average binding probablity and average logRatios

all motif sites

sites that are differentially open in hypoxia

sites that are differentially open in normoxia

Compare logRatios for differentially accessible sites using limma

Session information