loose rock

Set of Functions to Use in Survival Analysis and in Data Science

Collection of function to improve workflow in survival analysis and data science. Among the many features, the generation of balanced datasets, retrieval of protein coding genes from two public databases (live) and generation of random matrix based on covariance matrix.

The work has been mainly supported by two grants: FCT SFRH/BD/97415/2013 and the EU Commission under SOUND project with contract number 633974.

Install

The only pre-requirement is to install biomaRt bioconductor package as it cannot be installed automatically via CRAN.

All other dependencies should be installed when running the install command.

if (!require("BiocManager"))
  install.packages("BiocManager")
BiocManager::install("loose.rock")

# use the package
library(loose.rock)

Overview

• coding.genes(): downloads protein coding genes from external databases
• gen.synth.xdata(): generate random matrix with pre-determined covariance
• balanced.cv.folds() and balanced.train.and.test(): get balanced train/test sets and cv folds.
• run.cache(): keep cache or results of a function
• proper() : Capitalize string using regexpression
• my.colors() : My own pallete
• my.symbols() : Same with symbols to plots
• … check out rest of Documentation

Libraries required for this vignette

library(dplyr)

Get a current list of protein coding genes

Showing only a random sample of 15

coding.genes() %>%
  dplyr::arrange(external_gene_name) %>% {
   dplyr::slice(., sample(seq(nrow(.)), 15)) 
  } %>%
  knitr::kable()

ensembl_gene_id	external_gene_name
ENSG00000163626	COX18
ENSG00000197208	SLC22A4
ENSG00000115806	GORASP2
ENSG00000125414	MYH2
ENSG00000196453	ZNF777
ENSG00000082556	OPRK1
ENSG00000185475	TMEM179B
ENSG00000058272	PPP1R12A
ENSG00000132446	FTHL17
ENSG00000230268	SSU72P8
ENSG00000274540	PRR23D1
ENSG00000128342	LIF
ENSG00000196371	FUT4
ENSG00000103932	RPAP1
ENSG00000123473	STIL

Balanced test/train dataset

This is specially relevant in survival or binary output with few cases of one category that need to be well distributed among test/train datasets or in cross-validation folds.

Example below sets aside 90% of the data to the training set. As samples are already divided in two sets (set1 and set2), it performs the 90% separation for each and then joins (with option join.all = T) the result.

set1 <- c(rep(TRUE, 8), FALSE, rep(TRUE, 9), FALSE, TRUE)
set2 <- !set1
cat(
  'Set1', '\n', set1, '\n\n',
  'Set2', '\n', set2, '\n\n',
  'Training / Test set using logical indices', '\n\n'
)
set.seed(1985)
balanced.train.and.test(set1, set2, train.perc = .9)
#
set1 <- which(set1)
set2 <- which(set2)
cat(
  '##### Same sets but using numeric indices', '\n\n', 
  'Set1', '\n', set1, '\n\n', 
  'Set2', '\n', set2, '\n\n', 
  'Training / Test set using numeric indices', '\n')
set.seed(1985)
balanced.train.and.test(set1, set2, train.perc = .9)
#
#> Set1 
#>  TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE FALSE TRUE 
#> 
#>  Set2 
#>  FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE TRUE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE TRUE FALSE 
#> 
#>  Training / Test set using logical indices 
#> 
#> $train
#>  [1]  1  2  3  4  5  6  7  8  9 10 11 12 14 15 17 18 20
#> 
#> $test
#> [1] 13 16 19
#> 
#> ##### Same sets but using numeric indices 
#> 
#>  Set1 
#>  1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18 20 
#> 
#>  Set2 
#>  9 19 
#> 
#>  Training / Test set using numeric indices 
#> $train
#>  [1]  1  2  3  4  5  6  7  8  9 10 11 12 14 15 17 18 20
#> 
#> $test
#> [1] 13 16 19

Generate synthetic matrix with covariance

xdata1 <- gen.synth.xdata(10, 5, .2)
xdata2 <- gen.synth.xdata(10, 5, .75)

#> Using .2^|i-j| to generate co-variance matrix
#> X generated
#> cov(X)

#> Using .75^|i-j| to generate co-variance matrix (plotting correlation)
#> X generated
#> cov(X)

Save in cache

Uses a cache to save and retrieve results. The cache is automatically created with the arguments and source code for function, so that if any of those changes, the cache is regenerated.

Caution: Files are not deleted so the cache directory can become rather big.

Set a temporary directory to save all caches (optional)

base.dir(file.path(tempdir(), 'run-cache'))
#> [1] "/tmp/Rtmpd4sOlL/run-cache"

Run sum function twice

a <- run.cache(sum, 1, 2)
#> Saving in cache:  /tmp/Rtmpd4sOlL/run-cache/8ca6/cache-generic_cache-H_8ca697a81d8184a82de72523a678a4290375a07e304dd20a78bd488827978af3.RData
b <- run.cache(sum, 1, 2)
#> Loading from cache (not calculating):
#>   /tmp/Rtmpd4sOlL/run-cache/8ca6/cache-generic_cache-H_8ca697a81d8184a82de72523a678a4290375a07e304dd20a78bd488827978af3.RData
#> Cache was created at 2021-03-15 16:24:17 using loose.rock v1.1.2
all(a == b)
#> [1] TRUE

Run rnorm function with an explicit seed (otherwise it would return the same random number)

a <- run.cache(rnorm, 5, seed = 1985)
#> Saving in cache:  /tmp/Rtmpd4sOlL/run-cache/9fda/cache-generic_cache-H_9fdab5baa36653c6d435ce2d68ec6651845f679861f463fe065f38115dc7acbe.RData
b <- run.cache(rnorm, 5, seed = 2000)
#> Saving in cache:  /tmp/Rtmpd4sOlL/run-cache/2ada/cache-generic_cache-H_2adac402358921459b509ec972477640ce54df8436844fb57f761cbe49a3296d.RData
all(a == b)
#> [1] FALSE

Proper

One of such is a proper function that capitalizes a string.

x <- "OnE oF sUcH iS a proPer function that capitalizes a string."
proper(x)
#> [1] "One Of Such Is A Proper Function That Capitalizes A String."

Custom colors and symbols

my.colors() and my.symbols() can be used to improve plot readability.

xdata <- -10:10
plot(
  xdata, 1/10 * xdata * xdata + 1, type="l", 
  pch = my.symbols(1), col = my.colors(1), cex = .9,
  xlab = '', ylab = '', ylim = c(0, 20)
)
grid(NULL, NULL, lwd = 2) # grid only in y-direction
for (ix in 2:22) {
  points(
    xdata, 1/10 * xdata * xdata + ix, pch = my.symbols(ix), 
    col = my.colors(ix), cex = .9
  )
}