Sort anyting to anything

Here is how a classical .pairs file looks like:

For more details about .pairs file format please look at the documentation

To sort pairs in either the .pairs or .parquet format, you just need to run a simple CLI command (with parameters similar to those used in pairtools). Note that sorting can be combined with conversion - and in most cases, you’ll actually save time by combining processing(in this case sorting) and conversion, rather than performing them separately.

Here’s an example with some useful parameters:

pairs_to_parquet sort -o test_sorted.parquet --nproc 4 --tmpdir /temp_directory --memory 35G test.pairs.gz

Default option of sorting is to sort pairs by chr1-chr2-pos1-pos2-pair_type, where chr and pair_type columns are sorted in lexicographical order and pos columns in numerical order

If you have your own idea, in which order you want to sort you pairs: then you need to specify c1, c2, p1, p2, pt or even extra_col parameters