For the following data, if the p.2 values is greater than 0.5, I would like to swap the values of column-1(b.1.1) and column-2 b.1.2,
column-3(b.2.1) and column-4 b.2.2 , column-5(b.3.1) and column-6 b.3.2
mydata
b.1.1 b.1.2 b.2.1 b.2.2 b.3.1 b.3.2 p.1 p.2
1 0.40772028 0.43064446 0.2697412 0.9191535 0.1523922 0.7629324 0.86061981 0.1393802
2 -0.77459375 0.79860856 -0.5263932 -0.5640303 0.5131236 0.6472614 0.63494425 0.3650557
3 -0.06088828 0.42685669 -1.0643744 0.8330836 0.1184059 0.6661079 0.07382585 0.9261742
4 1.54204242 -0.08987067 -0.7365012 0.3762336 0.3781115 -0.7340340 0.65481949 0.3451805
5 -0.73397310 1.34927693 0.2202689 0.2422944 1.5267535 -0.5207967 0.54425551 0.4557445
For example, in the first row, between p.1 and p.2, p.2 is not greater than 0.5, I am not swapping any values of this row. In the third row, p.2 is greater than 0.5, So, I would like to swap the row values of each beta pair written above.
Any help is appreciated.
Here is an option. Create an index to subset the columns that starts with 'b' ('i1'), and another index to subset the rows ('i2'). Then, split the dataset into a list of datasets based on the similarity of column names after subsetting the rows ('i2'), loop through the list, reverse the elements, cbind the list of data.frames and update that to the original dataset rows/columns
i1 <- startsWith(names(mydata), "b")
i2 <- mydata$p.2 > 0.5
mydata[i2, i1] <- do.call(cbind,
lapply(split.default(mydata[i2, i1, drop = FALSE],
sub("\\.\\d+$", "", names(mydata)[i1])), rev))
mydata
# b.1.1 b.1.2 b.2.1 b.2.2 b.3.1 b.3.2 p.1 p.2
#1 0.4077203 0.43064446 0.2697412 0.9191535 0.1523922 0.7629324 0.86061981 0.1393802
#2 -0.7745937 0.79860856 -0.5263932 -0.5640303 0.5131236 0.6472614 0.63494425 0.3650557
#3 0.4268567 -0.06088828 0.8330836 -1.0643744 0.6661079 0.1184059 0.07382585 0.9261742
#4 1.5420424 -0.08987067 -0.7365012 0.3762336 0.3781115 -0.7340340 0.65481949 0.3451805
#5 -0.7339731 1.34927693 0.2202689 0.2422944 1.5267535 -0.5207967 0.54425551 0.4557445
Another option is tidyverse, where we convert into 'long' format, do the transformation in that form, and reshape back to 'wide' format
library(dplyr)
library(tidyr)
library(stringr)
library(tibble)
mydata %>%
rownames_to_column('rn') %>%
pivot_longer(cols = -c(rn, p.1, p.2)) %>%
group_by(rn, grp = str_remove(name, "\\.\\d+$")) %>%
mutate(value = case_when(p.2 > 0.5 ~ rev(value), TRUE ~ value)) %>%
ungroup %>%
select(-grp) %>%
pivot_wider(names_from = name, values_from = value) %>%
select(names(mydata))
# A tibble: 5 x 8
# b.1.1 b.1.2 b.2.1 b.2.2 b.3.1 b.3.2 p.1 p.2
# <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#1 0.408 0.431 0.270 0.919 0.152 0.763 0.861 0.139
#2 -0.775 0.799 -0.526 -0.564 0.513 0.647 0.635 0.365
#3 0.427 -0.0609 0.833 -1.06 0.666 0.118 0.0738 0.926
#4 1.54 -0.0899 -0.737 0.376 0.378 -0.734 0.655 0.345
#5 -0.734 1.35 0.220 0.242 1.53 -0.521 0.544 0.456
mydata <- structure(list(b.1.1 = c(0.40772028, -0.77459375, -0.06088828,
1.54204242, -0.7339731), b.1.2 = c(0.43064446, 0.79860856, 0.42685669,
-0.08987067, 1.34927693), b.2.1 = c(0.2697412, -0.5263932, -1.0643744,
-0.7365012, 0.2202689), b.2.2 = c(0.9191535, -0.5640303, 0.8330836,
0.3762336, 0.2422944), b.3.1 = c(0.1523922, 0.5131236, 0.1184059,
0.3781115, 1.5267535), b.3.2 = c(0.7629324, 0.6472614, 0.6661079,
-0.734034, -0.5207967), p.1 = c(0.86061981, 0.63494425, 0.07382585,
0.65481949, 0.54425551), p.2 = c(0.1393802, 0.3650557, 0.9261742,
0.3451805, 0.4557445)), class = "data.frame", row.names = c("1",
"2", "3", "4", "5"))