Using pipes

Concatenating commands in pipes

KGTK has a pipelining architecture based on Unix pipes that allows chaining all the commands included in the framework.

Pipelining increases efficiency by avoiding the need to write files to disk and supporting parallelism, allowing downstream commands to process data before upstream commands complete.

To use pipes, you just need to use the shortcut pipe operator /. For example, the following command calculates the number of instances of each class in a KGTK file edges.tsv by first filtering all P31 (instance of) edges and then counting them with the unique command. The result is stored in a file named class_count.tsv:

kgtk filter -i edges.tsv -p " ; P31 ; " / unique --column node2 -o class_count.tsv

Note

If you need to read from stdin you will need to use -i -