Filter data.frame rows by a logical condition

Question

I want to filter rows from a data.frame based on a logical condition. Let\'s suppose that I have data frame like

   expr_value     cell_type
1           


        

Answer 1

To select rows according to one 'cell_type' (e.g. 'hesc'), use ==:

expr[expr$cell_type == "hesc", ]

To select rows according to two or more different 'cell_type', (e.g. either 'hesc' or 'bj fibroblast'), use %in%:

expr[expr$cell_type %in% c("hesc", "bj fibroblast"), ]

Answer 2

we can use data.table library

  library(data.table)
  expr <- data.table(expr)
  expr[cell_type == "hesc"]
  expr[cell_type %in% c("hesc","fibroblast")]

or filter using %like% operator for pattern matching

 expr[cell_type %like% "hesc"|cell_type %like% "fibroblast"]

Answer 3

Use subset (for interactive use)

subset(expr, cell_type == "hesc")
subset(expr, cell_type %in% c("bj fibroblast", "hesc"))

or better dplyr::filter()

filter(expr, cell_type %in% c("bj fibroblast", "hesc"))

Answer 4

You could use the dplyr package:

library(dplyr)
filter(expr, cell_type == "hesc")
filter(expr, cell_type == "hesc" | cell_type == "bj fibroblast")

Answer 5

No one seems to have included the which function. It can also prove useful for filtering.

expr[which(expr$cell == 'hesc'),]

This will also handle NAs and drop them from the resulting dataframe.

Running this on a 9840 by 24 dataframe 50000 times, it seems like the which method has a 60% faster run time than the %in% method.

Answer 6

Sometimes the column you want to filter may appear in a different position than column index 2 or have a variable name.

In this case, you can simply refer the column name you want to filter as:

columnNameToFilter = "cell_type"
expr[expr[[columnNameToFilter]] == "hesc", ]