core_commons.R 15 KB
Newer Older
Holger Brandl's avatar
Holger Brandl committed
1 2 3 4
########################################################################################################################
## set a default cran r mirror  and customize environment

#cat(".Rprofile: Setting Goettingen repository\n")
Holger Brandl's avatar
Holger Brandl committed
5 6
#todo consider to use chooseCRANmirror(graphics=FALSE, ind=10) instead

Holger Brandl's avatar
Holger Brandl committed
7 8 9 10 11 12 13 14 15 16 17 18
r = getOption("repos") # hard code the UK repo for CRAN

r["CRAN"] = "http://ftp5.gwdg.de/pub/misc/cran/"
options(repos = r)
rm(r)


## user browser for help
options(help_type="html")

## plot more characters per line
options(width=150)

19 20 21 22 23
## adjust dplyr printing settings
## http://stackoverflow.com/questions/22471256/overriding-variables-not-shown-in-dplyr-to-display-all-columns-from-df
## http://stackoverflow.com/questions/29396111/dplyrtbl-df-fill-whole-screen
options(dplyr.print_max = 20) ## num rows
options(dplyr.width = 130) ## width
Holger Brandl's avatar
Holger Brandl committed
24
#options(dplyr.width = 280); options(width=280) ## width
25

Holger Brandl's avatar
Holger Brandl committed
26 27 28
# for sqldf to avoid the use of tckl
options(gsubfn.engine = "R")

Holger Brandl's avatar
Holger Brandl committed
29

Holger Brandl's avatar
Holger Brandl committed
30
########################################################################################################################
Holger Brandl's avatar
Holger Brandl committed
31
## automatic package installation
Holger Brandl's avatar
Holger Brandl committed
32

Holger Brandl's avatar
Holger Brandl committed
33

34
## externalized installer to also allow for installation without loading
35
install_package <- function(x){
36 37 38 39 40 41 42 43 44 45 46 47 48 49
    if(!isTRUE(x %in% .packages(all.available=TRUE)) && any(available.packages()[,1]==x)) {
        # update.packages(ask=F) # update dependencies, if any.
        eval(parse(text=paste("install.packages('", x, "')", sep="")))
    }

    ## if it's still missing check if it's on bioconductor
    if(!isTRUE(x %in% .packages(all.available=TRUE))) {
        bcPackages <- as.vector(read.dcf(url("https://bioconductor.org/packages/3.3/bioc/src/contrib/PACKAGES"), "Package"))

        if(any(bcPackages==x)){
            source("http://bioconductor.org/biocLite.R")
            eval(parse(text=paste("biocLite('", x, "', ask=FALSE)", sep="")))
        }
    }
50 51
}

Holger Brandl's avatar
Holger Brandl committed
52 53 54
load_pack <-  function(x, warn_conflicts=T){
    x <- as.character(substitute(x));

55
   install_package(x)
Holger Brandl's avatar
Holger Brandl committed
56

57
    ## load it using a library function so that load_pack errors if package is still not ins
Holger Brandl's avatar
Holger Brandl committed
58 59 60
    eval(parse(text=paste("library(", x, ",  quietly=T, warn.conflicts=", warn_conflicts, ")", sep="")))
}

61

62 63 64 65 66 67
check_version = function(pkg_name, min_version) {
    cur_version = packageVersion(pkg_name)
    if(cur_version < min_version) stop(sprintf("Package %s needs a newer version,
               found %s, need at least %s", pkg_name, cur_version, min_version))
}
#check_version("dplyr", "0.4-1")
Holger Brandl's avatar
Holger Brandl committed
68

Holger Brandl's avatar
Holger Brandl committed
69
########################################################################################################################
Holger Brandl's avatar
Holger Brandl committed
70
## load core packages
Holger Brandl's avatar
Holger Brandl committed
71

Holger Brandl's avatar
Holger Brandl committed
72 73
#load_pack(plyr)
#load_pack(reshape2)
74
#load_pack(reshape2, quietly=T, warn_conflicts=F)
Holger Brandl's avatar
Holger Brandl committed
75

Holger Brandl's avatar
Holger Brandl committed
76
## load on purpose after plyr
77 78 79
load_pack(dplyr, warn_conflicts=F)
load_pack(magrittr, warn_conflicts=F)
load_pack(tidyr, warn_conflicts=F)
Holger Brandl's avatar
Holger Brandl committed
80 81 82
load_pack(stringr)
load_pack(purrr)
load_pack(readr)
Holger Brandl's avatar
Holger Brandl committed
83
load_pack(forcats)
Holger Brandl's avatar
Holger Brandl committed
84
load_pack(readxl) ## supress differring build number
85

Holger Brandl's avatar
Holger Brandl committed
86
## needed for caching
87
load_pack(digest)
Holger Brandl's avatar
Holger Brandl committed
88 89 90

#suppressWarnings(load_pack(readxl)) ## supress differring build number

91
#load_pack(readxl) ## supress differring build number
92 93 94


## common plotting requirements since they are omnipresent
95 96 97
load_pack(ggplot2)
load_pack(scales, warn_conflicts=F)
load_pack(grid)
98

Holger Brandl's avatar
Holger Brandl committed
99
## for table exploration without using Rstudio
100
load_pack(DT)
Holger Brandl's avatar
Holger Brandl committed
101

Holger Brandl's avatar
Holger Brandl committed
102 103

## moved into datatable_commons because replaced almost everywhere with dplyr
104
#load_pack(data.table)
Holger Brandl's avatar
Holger Brandl committed
105 106 107 108 109




########################################################################################################################
Holger Brandl's avatar
Holger Brandl committed
110
#### Convenience aliases
Holger Brandl's avatar
Holger Brandl committed
111 112


Holger Brandl's avatar
Holger Brandl committed
113
echo <- function(...) cat(paste(...), fill=T)
Holger Brandl's avatar
Holger Brandl committed
114 115 116 117

ac <- function(...) as.character(...)

# string concatenation without space gaps (could/should use paste0 instead)
Holger Brandl's avatar
Holger Brandl committed
118 119
## Deprecated: use paste0 instead
#concat <- function(...) paste(..., sep="")
Holger Brandl's avatar
Holger Brandl committed
120 121 122 123 124

unlen <- function(x) length(unique(x))

pp <- function(dat) page(dat, method = "print")

Holger Brandl's avatar
Holger Brandl committed
125 126
as.df <- function(dt){ warning("DEPRECATED: use as_df instead of as.df"); as.data.frame(dt) }
as_df <- function(dt) as.data.frame(dt)
Holger Brandl's avatar
Holger Brandl committed
127

Holger Brandl's avatar
Holger Brandl committed
128

Holger Brandl's avatar
Holger Brandl committed
129 130 131 132 133
install_package("tibble")

## restore pre-tibble-v1.2 naming to creating data-frame in place
frame_data = function(...) tibble::tribble(...)

134 135 136

add_rownames = function(...) tibble::rownames_to_column(...)

Holger Brandl's avatar
Holger Brandl committed
137 138 139 140 141 142 143 144 145 146 147 148 149

## redefine dply::splat to allow for more streamlined rowwise df-processing
## see https://groups.google.com/forum/?utm_source=digest&utm_medium=email#!topic/manipulatr/E6WOcHlRJcw
#splat = function (flat) {
#    function(args, ...) {
#        do.call(flat, c(args, list(...)))
#    }
#}
## for now simply import just splat from plyr namespace
install_package("plyr")
splat = plyr::splat


Holger Brandl's avatar
Holger Brandl committed
150 151 152 153 154 155 156 157 158 159
########################################################################################################################
#### data.frame manipulation


shuffle <- function(df) df[sample(nrow(df)),]

first <- function(x, n=1) head(x,n)



Holger Brandl's avatar
Holger Brandl committed
160 161 162 163 164 165 166 167
vec2df <- function(namedVec){
    warning("DEPRECATED use vec_as_df instead of vec2df")
    namedVec %>% {data.frame(name=names(.), value=., row.names=NULL)}
}

vec_as_df <- function(namedVec, row_name="name", value_name="value"){
    data_frame(name=names(namedVec), value=namedVec) %>% set_names(row_name, value_name)
}
Holger Brandl's avatar
Holger Brandl committed
168

Holger Brandl's avatar
Holger Brandl committed
169 170

rownames2column <- function(df, colname){
171 172
    warning("DEPRECATED Use dplyr::add_rownames directly")
    add_rownames(df, var=colname)
Holger Brandl's avatar
Holger Brandl committed
173 174 175 176
}


column2rownames<- function(df, colname){
177
    #browser()
Holger Brandl's avatar
Holger Brandl committed
178
    ## force into df to avoid dplyr problems
Holger Brandl's avatar
Holger Brandl committed
179
    df <- as_df(df)
Holger Brandl's avatar
Holger Brandl committed
180

Holger Brandl's avatar
Holger Brandl committed
181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196
    rownames(df) <- ac(df[,colname])
    df[colname] <- NULL
    return(df)
}

## pushing some columns to the right end of a data.frame
push_right <- function(df, pushColNames){
    df[,c(setdiff(names(df),pushColNames), pushColNames)]
}


push_left <- function(df, pushColNames){
    df[,c(pushColNames, setdiff(names(df),pushColNames))]
}


Holger Brandl's avatar
Holger Brandl committed
197 198
#http://astrostatistics.psu.edu/datasets/R/html/base/html/formals.html
set_names <- function(df, ...){
199
    #browser()
Holger Brandl's avatar
Holger Brandl committed
200
    newnames <- as.character(unlist(list(...)))
201 202 203 204

    ## automatically convert matrices to data.frames (otherwise the names set would fail
    if(is.matrix(df)) df %<>% as.data.frame()

205
    names(df) <- newnames;
Holger Brandl's avatar
Holger Brandl committed
206 207
    return(df)
}
Holger Brandl's avatar
Holger Brandl committed
208 209
#iris %>% set_names(c("setosa", "hallo")) %>% head
#iris %>% set_names("setosa", "hallo") %>% head
Holger Brandl's avatar
Holger Brandl committed
210 211


212 213
rify_names  <- function(df){
    warning("DEPRECATED: use pretty_columns instead");
Holger Brandl's avatar
Holger Brandl committed
214
    names(df) <- names(df) %>% str_replace_all(c(" "="_", "-"="_", "/"="_"));
215 216 217
    df
}

218 219
pretty_columns  <- function(df){
    names(df) <- names(df) %>%
Holger Brandl's avatar
Holger Brandl committed
220
        str_replace_all("[#=.()/ -]+", "_") %>%
221 222 223 224 225
        str_replace("[_]+$", "") %>%
        str_replace("^[_]+", "") %>%
        tolower;
    df
}
Holger Brandl's avatar
Holger Brandl committed
226

227 228
# http://stackoverflow.com/questions/23188900/view-entire-dataframe-when-wrapped-in-tbl-df
print_all <- function(df) df %>% tbl_df %>% print(n = nrow(.))
Holger Brandl's avatar
Holger Brandl committed
229

Holger Brandl's avatar
Holger Brandl committed
230 231
head_html <- function(df, n=5) head(df, n) %>% knitr::kable(format="html") %>% print()

Holger Brandl's avatar
Holger Brandl committed
232 233
print_head <- function(df, desc=NULL){
    print(head(df))
Holger Brandl's avatar
Holger Brandl committed
234
    print(nrow(df))
Holger Brandl's avatar
Holger Brandl committed
235 236 237 238 239
    return(df)
}


fac2char <- function(mydata, convert=names(mydata)[sapply(mydata, is.factor)]){
Holger Brandl's avatar
Holger Brandl committed
240 241 242
    if(length(convert)==0){
        return(mydata)
    }
Holger Brandl's avatar
Holger Brandl committed
243 244 245 246 247 248 249 250 251 252 253 254 255

    inputColOrder <- names(mydata)

    convertData <- subset(mydata, select= names(mydata)%in%convert)
    convertData <- as.data.frame(lapply(convertData, as.character), stringsAsFactors = FALSE)

    keepData <-  subset(mydata, select=!(names(mydata)%in%convert))
    newdata <- cbind(convertData, keepData)
    newdata <- newdata[,inputColOrder]

    return(newdata)
}

Holger Brandl's avatar
Holger Brandl committed
256 257 258 259
## convenience method to sort factor levels with decreasing frequencies
fct_revfreq = function(x) fct_infreq(x) %>% fct_rev


Holger Brandl's avatar
Holger Brandl committed
260
## replace R within pipe change just use ... %>% do(replaceNA(0)) %>% ...
Holger Brandl's avatar
Holger Brandl committed
261
replaceNA <- function(x, withValue) { x[is.na(x)] <- withValue; x }
Holger Brandl's avatar
Holger Brandl committed
262

Holger Brandl's avatar
Holger Brandl committed
263

Holger Brandl's avatar
Holger Brandl committed
264

265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282

## see http://stackoverflow.com/questions/17288222/r-find-value-in-multiple-data-frame-columns/40586572#40586572
## case-insenstive search all columns of a data-frame with a fixed search term
search_df = function(df, search_term){
    apply(df, 1, function(r){
        any(str_detect(as.character(r), fixed(search_term, ignore_case=T)))
    }) %>% subset(df, .)
}


## filter a data-frame for those rows where at least one column is matching the given expression (that must evaluate to a boolean vector for each row).
match_df = function(df, search_expr){
    filter_fun = eval(substitute(function(x){search_expr}))

    apply(df, 1, function(r) any(filter_fun(r))) %>% subset(df, .)
}


Holger Brandl's avatar
Holger Brandl committed
283
## todo still needed since there's if_else in dplyr now
Holger Brandl's avatar
Holger Brandl committed
284
safe_ifelse <- function(cond, yes, no) {
285
    warning("DEPRECATED Use dplyr::if_else instead")
Holger Brandl's avatar
Holger Brandl committed
286

Holger Brandl's avatar
Holger Brandl committed
287 288 289 290 291 292
#browser()
    isfacOrChar <- function(x) class(x)  %in% c("factor", "character")

    if (isfacOrChar(yes) | isfacOrChar(no)) {
        yes <- ac(yes)
        no <- ac(no)
293
    }
Holger Brandl's avatar
Holger Brandl committed
294 295

    ifelse(cond,yes,no)
Holger Brandl's avatar
Holger Brandl committed
296 297
}

Holger Brandl's avatar
Holger Brandl committed
298

Holger Brandl's avatar
Holger Brandl committed
299
## for na instead use mutate_each with:
Holger Brandl's avatar
Holger Brandl committed
300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327
#empty_as_na <- function(x) safe_ifelse(x=="", NA, x)
#empty_as_na <- function(x) ifelse(class(x)  %in% c("factor", "character") & x=="", NA, x)
empty_as_na <- function(x){
    if("factor" %in% class(x)) x <- as.character(x) ## since ifelse wont work with factors
    ifelse(as.character(x)!="", x, NA)
}

#if(F){ ## DEBUG of empty_as_na
#cond <- allJobs %>% head %$% submit_time %>% c("")
#empty_as_na( cond)
#cond <- allJobs %>% head %$% resubmission_of
#empty_as_na( cond)
#
#empty_as_na( c(1, 2, NA))
#empty_as_na( c("sdf", "sdf2", NA))
#empty_as_na( c("sdf", "sdf2", ""))
#
#myFac <- as.factor(c("sdf", "sdf2", NA))
#empty_as_na( myFac)
#ifelse(as.character(myFac)!="", myFac, NA)
#
#empty_as_na( c("sdf", "sdf2", ""))
#
#iris[1,1] <- ""
#apply(iris, 2, function(x) gsub("^$|^ $", NA, x))
#}


Holger Brandl's avatar
Holger Brandl committed
328 329 330
## see http://stackoverflow.com/questions/24172111/change-the-blank-cells-to-na/33952598#33952598


Holger Brandl's avatar
Holger Brandl committed
331 332 333 334 335 336
## apply dplyr::filter to df but use filter criterions for cross-tabulation beforehand
filter_count <- function(df, ...){
    print(count(df, ...))
    filter(df, ...)
}

Holger Brandl's avatar
Holger Brandl committed
337 338 339 340 341 342
# Example: publications %>% count(journal) %>% as("num_pubs")
as = function(df, name){
    names(df)[length(names(df))] = name
    df
}

Holger Brandl's avatar
Holger Brandl committed
343

Holger Brandl's avatar
Holger Brandl committed
344 345
distinct_all = function (x, ...) distinct(x, ..., .keep_all=T)

Holger Brandl's avatar
Holger Brandl committed
346 347 348 349
## fetch a column of a matrix in a magrittr pipe. Useful along with str_*
get_col = function(data, col_index) data[, col_index] ## also could use magrittr::extract here


Holger Brandl's avatar
Holger Brandl committed
350

351 352 353 354 355 356 357
reload_dplyr <- function(){
    unloadNamespace('tidyr')
    unloadNamespace('dplyr')
    require(tidyr);require(dplyr)
}


358
## from http://stackoverflow.com/questions/7505547/detach-all-packages-while-working-in-r
359
unload_packages <- function() {
360 361 362 363 364
    basic.packages <- c("package:stats","package:graphics","package:grDevices","package:utils","package:datasets","package:methods","package:base")
    package.list <- search()[ifelse(unlist(gregexpr("package:",search()))==1,TRUE,FALSE)]
    package.list <- setdiff(package.list,basic.packages)
    if (length(package.list)>0)  for (package in package.list) detach(package, character.only=TRUE)
}
Holger Brandl's avatar
Holger Brandl committed
365 366


Holger Brandl's avatar
Holger Brandl committed
367 368 369 370
## workaround for biomart
## Deprecated: load dplyr after biomart to avoid this problem
#dselect <- function(...) dplyr::select(...)

Holger Brandl's avatar
Holger Brandl committed
371

Holger Brandl's avatar
Holger Brandl committed
372
########################################################################################################################
Holger Brandl's avatar
Holger Brandl committed
373
#### Result Caching for long running tasks
Holger Brandl's avatar
Holger Brandl committed
374

Holger Brandl's avatar
Holger Brandl committed
375 376
## related: http://cran.r-project.org/web/packages/R.cache/R.cache.pdf

377
cache_it <- function(expr, filePrefix="cache"){
378
    cacheFile <- paste0(filePrefix, "_", substr(digest::digest(deparse(expr)), 1,6)) %>% paste0(".", ., ".RData")
Holger Brandl's avatar
Holger Brandl committed
379

Holger Brandl's avatar
Holger Brandl committed
380 381 382 383 384 385 386 387 388 389 390
    if(file.exists(cacheFile)){
        local(get(load(cacheFile)))
    } else {
        result <- eval(expr)
        save(result, file=cacheFile)
        result
    }
}

## Examples
#mydata <- quote(iris %>% filter(Species=="setosa")) %>% cache_it("tt")
Holger Brandl's avatar
Holger Brandl committed
391 392
#mydata <- quote(iris %>% filter(Species=="setosa")) %>% cache_it()
#mydata <- quote( { print("evaluate expr"); iris %>% filter(Species=="setosa") } ) %>% cache_it()
Holger Brandl's avatar
Holger Brandl committed
393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410

########################################################################################################################
#### File System

is.directory <- function(dirname) !is.na(file.info(dirname)$isdir)


mcdir <- function(dirname){
    if(!file.exists(dirname)){
        dir.create(dirname)
    }

    setwd(dirname)
}

locload <- function(fileName) local(get(load(fileName)))


Holger Brandl's avatar
Holger Brandl committed
411
## tbd: it would be more efficient to use Reduce here (see http://stackoverflow.com/questions/34344214/how-to-join-multiple-data-frames-using-dplyr)
Holger Brandl's avatar
Holger Brandl committed
412 413 414 415 416 417 418 419 420 421
rmerge <- function(LDF, by, ...){
    DF <- LDF[[1]]
    for (i in 2:length(LDF)) {
        DF <- merge(DF, LDF[[i]], by=by)
    }
    DF
}



Holger Brandl's avatar
Holger Brandl committed
422 423
## Deprecated: use trim_ext instead
trimEnd <- function(fileNames, ...) trim_ext(fileNames, ...)
Holger Brandl's avatar
Holger Brandl committed
424

Holger Brandl's avatar
Holger Brandl committed
425
trim_ext <-function(fileNames, exts=c()){
Holger Brandl's avatar
Holger Brandl committed
426 427 428 429 430 431 432
    for(fileExt in exts){
        fileNames <- str_replace(fileNames, paste(fileExt, "$",sep=""), "")
    }

    fileNames
}

Holger Brandl's avatar
Holger Brandl committed
433
## DEPRECATED Use write_tsv instead
Holger Brandl's avatar
Holger Brandl committed
434 435
write.delim <- function(df, file, header=TRUE,...){
    write.table(df, file, row.names=FALSE, col.names=header, sep="\t", ...)
Holger Brandl's avatar
Holger Brandl committed
436 437 438 439 440 441 442 443 444 445
}

rmSomeElements <- function(vec, toDel) vec[!(vec %in% toDel)]

rmLastElement <- function(vec) vec[-length(vec)]


########################################################################################################################
## Memory management

Holger Brandl's avatar
Holger Brandl committed
446

Holger Brandl's avatar
Holger Brandl committed
447
# improved list of objects
Holger Brandl's avatar
Holger Brandl committed
448 449
lsos <- function (pos = 1, pattern, order.by, decreasing=FALSE, head=FALSE, n=5) {
    napply <- function(names, fn) sapply(names, function(x) fn(get(x, pos = pos)))
Holger Brandl's avatar
Holger Brandl committed
450
    names <- ls(pos = pos, pattern = pattern)
Holger Brandl's avatar
Holger Brandl committed
451

Holger Brandl's avatar
Holger Brandl committed
452 453 454 455 456
    obj.class <- napply(names, function(x) as.character(class(x))[1])
    obj.mode <- napply(names, mode)
    obj.type <- ifelse(is.na(obj.class), obj.mode, obj.class)
    obj.size <- napply(names, object.size)/1000000
    obj.dim <- t(napply(names, function(x)
Holger Brandl's avatar
Holger Brandl committed
457

458
    as.numeric(dim(x))[1:2]))
Holger Brandl's avatar
Holger Brandl committed
459 460
    vec <- is.na(obj.dim)[, 1] & (obj.type != "function")
    obj.dim[vec, 1] <- napply(names, length)[vec]
Holger Brandl's avatar
Holger Brandl committed
461

Holger Brandl's avatar
Holger Brandl committed
462 463
    out <- data.frame(obj.type, obj.size, obj.dim)
    names(out) <- c("Type", "Size", "Rows", "Columns")
Holger Brandl's avatar
Holger Brandl committed
464

Holger Brandl's avatar
Holger Brandl committed
465
    if (!missing(order.by))
466
    out <- out[order(out[[order.by]], decreasing=decreasing), ]
Holger Brandl's avatar
Holger Brandl committed
467 468 469

    if (head) out <- head(out, n)

Holger Brandl's avatar
Holger Brandl committed
470 471 472 473 474
    out <- transform(out, var_name=rownames(out))
    rownames(out) <- NULL
    arrange(out, Size)
}

Holger Brandl's avatar
Holger Brandl committed
475
# shorthand that just shows top 1 results
Holger Brandl's avatar
Holger Brandl committed
476 477 478 479
lsosh <- function(..., n=10) {
    lsos(..., order.by="Size", decreasing=TRUE, head=TRUE, n=n)
}

Holger Brandl's avatar
Holger Brandl committed
480 481 482 483 484 485 486 487 488 489
########################################################################################################################
### Statistics


## outlier handling
trim_outliers <- function(values, range=quantile(values, c(0.05, 0.95)))  pmax(range[1], pmin(range[2], values))

## use trim_outliers instead
#limit_range <- function(values, range)  pmax(range[1], pmin(range[2], values))

Holger Brandl's avatar
Holger Brandl committed
490
se <- function(x) sd(x, na.rm=TRUE) / sqrt(sum(!is.na(x)))
491 492 493 494 495 496 497 498


########################################################################################################################
### Misc

## inspired by http://stackoverflow.com/questions/8343509/better-error-message-for-stopifnot
assert <- function (expr, error) {
    if (! expr) stop(error, call. = FALSE)
499 500 501
}

### table rendering
Holger Brandl's avatar
Holger Brandl committed
502 503
table_browser <- function(df, caption=deparse(substitute(df)), ...){
    datatable(df, filter = "bottom", extensions = 'Buttons', options = list( dom = 'Bfrtip', buttons = c('copy', 'csv', 'excel')), caption=caption,...)
504
}