add files

compile-report.R

@@ -0,0 +1,14 @@
+#!/usr/bin/env Rscript
+
+## USAGE: compile_RMD_report.R report.Rmd
+## Requires pandoc version 1.13+
+# module load pandoc/1.13.1
+
+# ~~~~~ GET SCRIPT ARGS ~~~~~~~ #
+args <- commandArgs(TRUE)
+input_dir <- args[1]
+
+
+#  ~~~~~ COMPILE ~~~~~ #
+Rmdfile <- "peaks-report.Rmd"
+rmarkdown::render(input = Rmdfile, params = list(input_dir = input_dir))

peaks-report.Rmd

@@ -0,0 +1,108 @@
+---
+title: "Peaks per Sample Report"
+author: "`r system('whoami', intern = TRUE)`"
+date: "`r format(Sys.time(), '%B %d, %Y')`"
+output:
+  html_document:
+    keep_md: true
+    df_print: paged
+params:
+   input_dir: "/path/to/peaks_dir"
+---
+
+```{r setup, include=FALSE}
+# ~~~~~ CHUNK OPTIONS ~~~~~ # 
+knitr::opts_chunk$set(echo = FALSE, warning = FALSE)
+
+# ~~~~~ LIBRARIES ~~~~~ # 
+library("ggplot2")
+library("reshape2")
+library("DT")
+library("scales")
+
+# ~~~~~ FUNCTIONS ~~~~~ # 
+find_all_beds <- function (input_dirs, name_pattern = FALSE) {
+    # find all .bed files in the supplied dirs
+    bed_files <- dir(input_dirs, pattern = '.bed', full.names = TRUE, recursive = TRUE)
+    if(name_pattern != FALSE) bed_files <- bed_files[which(basename(bed_files) %in% as.character(name_pattern))]
+    return(bed_files)
+}
+
+get_numlines <- function(input_file) {
+    # count the number of lines in a file
+    return(length(readLines(input_file)))
+}
+
+mycat <- function(text){
+    # function for formatting text in the report
+    cat(gsub(pattern = "\n", replacement = "  \n", x = text))
+}
+
+split_sampleID_col <- function(df, new_colnames = NA, sample_colname = "sample", split_char = "-"){
+    # split the 'sample' column in the df into separate columns, return the new df with all columns
+    split_df <- as.data.frame(do.call(rbind, strsplit(x = df[[sample_colname]], split = split_char)))
+    if (! is.na(new_colnames)) colnames(split_df) <- new_colnames
+    df <- cbind(split_df, df)
+    return(df)
+}
+
+make_bed_df <- function(input_dir){
+    # create a dataframe from the bed files found in the dir
+    bed_files <- find_all_beds(input_dirs = input_dir, name_pattern = "peaks.bed") 
+    
+    # sample ID is saved in the dirname of each bed file
+    samples <- basename(dirname(bed_files))
+    
+    bed_df <- data.frame(sample = basename(dirname(bed_files)), 
+                         peaks = sapply(X = bed_files, FUN = get_numlines), 
+                         # file = bed_files, 
+                         stringsAsFactors = FALSE)
+    
+    # split the sample ID's into separate columns
+    bed_df <- split_sampleID_col(df = bed_df, new_colnames = c("patient", "status", "mark")) # ABC-R-H3K9AC
+    
+    # reindex the rownames
+    rownames(bed_df) <- seq(nrow(bed_df))
+    return(bed_df)
+}
+
+make_peaks_barplot <- function(df){
+    # make the ggplot bar plot
+    
+    # re order the df
+    # bed_df[order(bed_df[["peaks"]], decreasing = TRUE), ]
+    
+    peaks_barplot <- ggplot(data = df, aes(x = sample, y = peaks)) + geom_bar(stat = 'identity') + coord_flip() + ggtitle("Peaks per Sample") + scale_y_continuous(labels = comma)
+    
+    return(peaks_barplot)
+}
+
+
+# ~~~~~ GLOBALS ~~~~~ #
+input_dir <- normalizePath(params$input_dir)
+
+bed_df <- make_bed_df(input_dir = input_dir)
+
+peaks_barplot <- make_peaks_barplot(df = bed_df)
+```
+
+# Input dir
+
+```{r}
+mycat(input_dir)
+```
+
+# Peaks per Sample
+
+## Plot
+
+```{r, fig.height = 18}
+peaks_barplot
+```
+
+## Table
+
+```{r}
+datatable(data = bed_df, rownames = FALSE, options = list(pageLength = -1))
+```
+

test/Sample1-D-H3K27AC/peaks.bed

@@ -0,0 +1,100 @@
+chr1	713404	714771	PEAK_000000006879	5.43218
+chr1	762076	763696	PEAK_000000009942	4.73610
+chr1	805301	805587	PEAK_000000018007	2.99584
+chr1	839714	842477	PEAK_000000015416	3.27085
+chr1	856369	856864	PEAK_000000034513	2.52153
+chr1	858393	861419	PEAK_000000013148	3.77847
+chr1	875351	878268	PEAK_000000015147	3.31411
+chr1	893645	896889	PEAK_000000003419	6.19332
+chr1	901689	902860	PEAK_000000010313	4.63175
+chr1	934368	937436	PEAK_000000017213	3.05593
+chr1	948124	949646	PEAK_000000009603	4.82685
+chr1	954523	956380	PEAK_000000000880	7.14794
+chr1	968147	969541	PEAK_000000014654	3.41174
+chr1	974418	976164	PEAK_000000037159	2.48553
+chr1	994390	995547	PEAK_000000002158	6.54421
+chr1	1003650	1006104	PEAK_000000011767	4.20425
+chr1	1050525	1052368	PEAK_000000006930	5.42175
+chr1	1072282	1073285	PEAK_000000014210	3.50851
+chr1	1092848	1094301	PEAK_000000006902	5.42760
+chr1	1105587	1107720	PEAK_000000042428	2.42152
+chr1	1136244	1137052	PEAK_000000010470	4.58344
+chr1	1140653	1142169	PEAK_000000031526	2.56725
+chr1	1143840	1145034	PEAK_000000032154	2.55792
+chr1	1147654	1149392	PEAK_000000021174	2.82053
+chr1	1166113	1168342	PEAK_000000001375	6.85551
+chr1	1208078	1209925	PEAK_000000000123	8.23012
+chr1	1240557	1244841	PEAK_000000008691	5.04208
+chr1	1259462	1260984	PEAK_000000007869	5.22741
+chr1	1283563	1285374	PEAK_000000000063	8.58599
+chr1	1288279	1288515	PEAK_000000032162	2.55787
+chr1	1306160	1307980	PEAK_000000037146	2.48562
+chr1	1309853	1311572	PEAK_000000005586	5.69834
+chr1	1333342	1336045	PEAK_000000004572	5.92366
+chr1	1341487	1343777	PEAK_000000003950	6.06451
+chr1	1368998	1371808	PEAK_000000014547	3.43596
+chr1	1373100	1374030	PEAK_000000040526	2.44325
+chr1	1406334	1408471	PEAK_000000004957	5.83228
+chr1	1440626	1440846	PEAK_000000028336	2.62401
+chr1	1446478	1448450	PEAK_000000006910	5.42488
+chr1	1475465	1476360	PEAK_000000005356	5.74826
+chr1	1508757	1511179	PEAK_000000002165	6.54159
+chr1	1534959	1535876	PEAK_000000014091	3.53432
+chr1	1549467	1552067	PEAK_000000000679	7.30268
+chr1	1564643	1565080	PEAK_000000049659	2.34029
+chr1	1565949	1567416	PEAK_000000015905	3.20072
+chr1	1589876	1591162	PEAK_000000012383	4.01229
+chr1	1609034	1609548	PEAK_000000035140	2.51249
+chr1	1623066	1624605	PEAK_000000013084	3.80028
+chr1	1655486	1655994	PEAK_000000013468	3.69653
+chr1	1677089	1677831	PEAK_000000015922	3.19824
+chr1	1702846	1703786	PEAK_000000049108	2.34671
+chr1	1708526	1712270	PEAK_000000007659	5.27064
+chr1	1714036	1714653	PEAK_000000040436	2.44420
+chr1	1765283	1766236	PEAK_000000052526	2.29971
+chr1	1772925	1773268	PEAK_000000051732	2.31138
+chr1	1792039	1792938	PEAK_000000045313	2.38872
+chr1	1819973	1823412	PEAK_000000000022	8.96714
+chr1	1839750	1841467	PEAK_000000010323	4.62888
+chr1	1849915	1850748	PEAK_000000010575	4.55140
+chr1	1950860	1951292	PEAK_000000016615	3.11655
+chr1	1956661	1957971	PEAK_000000038894	2.46306
+chr1	1959060	1960527	PEAK_000000021343	2.81364
+chr1	1976148	1977208	PEAK_000000041117	2.43639
+chr1	1981318	1981581	PEAK_000000041520	2.43204
+chr1	2064258	2065186	PEAK_000000022206	2.78181
+chr1	2120628	2121651	PEAK_000000011314	4.34242
+chr1	2125581	2126838	PEAK_000000003933	6.06731
+chr1	2135908	2137180	PEAK_000000016024	3.18550
+chr1	2143863	2144767	PEAK_000000015608	3.24085
+chr1	2158552	2161581	PEAK_000000012689	3.92487
+chr1	2165597	2165828	PEAK_000000043032	2.41447
+chr1	2186803	2187637	PEAK_000000048717	2.35126
+chr1	2221231	2224317	PEAK_000000031089	2.57444
+chr1	2225406	2226346	PEAK_000000032541	2.55142
+chr1	2230994	2233972	PEAK_000000021175	2.82051
+chr1	2245921	2247678	PEAK_000000009078	4.95317
+chr1	2292273	2293280	PEAK_000000048268	2.35298
+chr1	2294703	2295755	PEAK_000000019171	2.91808
+chr1	2305614	2307110	PEAK_000000026298	2.66705
+chr1	2310710	2311453	PEAK_000000040564	2.44298
+chr1	2318794	2324171	PEAK_000000010442	4.59511
+chr1	2343418	2344951	PEAK_000000008491	5.08851
+chr1	2456670	2458588	PEAK_000000002370	6.48110
+chr1	2477809	2480927	PEAK_000000015120	3.31970
+chr1	2486308	2488901	PEAK_000000011418	4.31155
+chr1	2509416	2510137	PEAK_000000014105	3.52990
+chr1	2516463	2518929	PEAK_000000008687	5.04303
+chr1	2573750	2575204	PEAK_000000001506	6.79734
+chr1	2585412	2585829	PEAK_000000046217	2.37794
+chr1	2986302	2987201	PEAK_000000024021	2.72394
+chr1	3370911	3371822	PEAK_000000010589	4.54649
+chr1	3433162	3433460	PEAK_000000032872	2.54587
+chr1	3446869	3448453	PEAK_000000007042	5.39841
+chr1	3454635	3455174	PEAK_000000046434	2.37530
+chr1	3486714	3487406	PEAK_000000028502	2.62084
+chr1	3491633	3492766	PEAK_000000026382	2.66486
+chr1	3495003	3498181	PEAK_000000038069	2.47337
+chr1	3509826	3510537	PEAK_000000016141	3.17296
+chr1	3513670	3515269	PEAK_000000018382	2.96975
+chr1	3525280	3528673	PEAK_000000013869	3.58733