mirror of
https://github.com/johrpan/ubigen.git
synced 2025-10-26 19:57:24 +01:00
Elias Projahn
cc17c13888
This commit also changes the way the standard deviation and mean expression are computed in general. Now, samples where the gene is not expressed at all are excluded before the computation.
60 lines
2.4 KiB
R
60 lines
2.4 KiB
R
# This scripts reads the input data (See input.R) and performs various
|
|
# computations on it in order to later use the results for computating scores
|
|
# for ubuiquitously expressed genes.
|
|
|
|
library(data.table)
|
|
library(here)
|
|
|
|
i_am("scripts/input.R")
|
|
|
|
data <- fread(here("scripts", "input", "data_long.csv.gz"))
|
|
|
|
data[, `:=`(
|
|
expression_median = median(expression),
|
|
expression_95 = quantile(expression, probs = 0.95)
|
|
), by = sample]
|
|
|
|
# Transform the expression logarithmically. The samples that don't express a
|
|
# gene at all will be left out intentionally.
|
|
data[expression > 0, expression_log := log2(expression)]
|
|
|
|
results <- data[, .(
|
|
median_expression = median(expression[expression > 0]),
|
|
iqr_expression = IQR(expression[expression > 0]),
|
|
mean_expression = mean(expression[expression > 0]),
|
|
sd_expression = sd(expression[expression > 0]),
|
|
median_expression_normalized = median(expression_log, na.rm = TRUE),
|
|
iqr_expression_normalized = IQR(expression_log, na.rm = TRUE),
|
|
mean_expression_normalized = mean(expression_log, na.rm = TRUE),
|
|
sd_expression_normalized = sd(expression_log, na.rm = TRUE),
|
|
above_zero = mean(expression > 0.0),
|
|
above_threshold = mean(expression > 50.0),
|
|
above_median = mean(expression > expression_median),
|
|
above_95 = mean(expression > expression_95)
|
|
), by = "gene"]
|
|
|
|
# Normalize values to the range of 0.0 to 1.0.
|
|
results[, `:=`(
|
|
median_expression_normalized =
|
|
(median_expression_normalized -
|
|
min(median_expression_normalized, na.rm = TRUE)) /
|
|
(max(median_expression_normalized, na.rm = TRUE) -
|
|
min(median_expression_normalized, na.rm = TRUE)),
|
|
iqr_expression_normalized =
|
|
(iqr_expression_normalized -
|
|
min(iqr_expression_normalized, na.rm = TRUE)) /
|
|
(max(iqr_expression_normalized, na.rm = TRUE) -
|
|
min(iqr_expression_normalized, na.rm = TRUE)),
|
|
mean_expression_normalized =
|
|
(mean_expression_normalized -
|
|
min(mean_expression_normalized, na.rm = TRUE)) /
|
|
(max(mean_expression_normalized, na.rm = TRUE) -
|
|
min(mean_expression_normalized, na.rm = TRUE)),
|
|
sd_expression_normalized =
|
|
(sd_expression_normalized -
|
|
min(sd_expression_normalized, na.rm = TRUE)) /
|
|
(max(sd_expression_normalized, na.rm = TRUE) -
|
|
min(sd_expression_normalized, na.rm = TRUE))
|
|
)]
|
|
|
|
fwrite(results, file = here("scripts", "output", "results.csv"))
|