This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Keep rows with different signs in two columns

Hi all, I have just a very simple question about subsetting data.

I have a data frame with many columns, two columns are indicating the fold change after RNA-seq data analysis. This two columns contain positive and negative values. I would like to keep the opposite values (negative in one column and positive in the other and vice versa) while removing those which are the same sign (negative and negative in the two columns or positive and positive).

Example:

df$1   df$2
 1.      2
-1.      1 
-1.      4
 3.     -2
 6.     -4
-6.     -6
 6.      1
-3.      7
......

 Resulting dataframe :

df$1.  df$2

-1.     1
-1.     4
 3.    -2
 6.    -4
-3.     7
......

Any help, highly appreciated, thanks!

r subset
• 4,155 views • link
updated by zx8754 zx8754 • written by paolo002 paolo002

Thanks a lot. It worked nicely.

• 1 views • link
written by paolo002 paolo002

2 answers

You can define a new column:

df$3 = df$1 * df$2

then

df <- subset(df, df$3 < 0, select = 1:2)
• 1 views • link
updated by zx8754 zx8754 • written by ZZzzzzhong ZZzzzzhong

Very nice, we could avoid creating a new column:

df1[ df1$col1 * df1$col2 < 0, ]
• 1 views • link
written by zx8754 zx8754

yeah, nice, then it is possible to avoid creating the additional column in the data frame, thanks!

• 1 views • link
written by paolo002 paolo002

Check the columns signs then filter, try this example:

# reproducible example data
df1 <- read.table(text = "col1 col2
1. 2
-1. 1 
-1. 4
3. -2
6. -4
-6. -6
6. 1
-3. 7", header = TRUE)

# check if columns have the same sign, then filter
df1[ !(sign(df1$col1) == sign(df1$col2)), ]

#   col1 col2
# 2   -1    1
# 3   -1    4
# 4    3   -2
# 5    6   -4
# 8   -3    7
• 1 views • link
written by zx8754 zx8754

nice, this solution is very good, it checks the sign without doing any calculation on it, that was what I was looking for, no need of calculation and it does not create any additional column, thanks a lot!

• 1 views • link
written by paolo002 paolo002

Log in to answer this question.

More posts like this