feat: sort and compress final pairs with pairtools

Dockerfile

@@ -17,7 +17,8 @@ RUN conda install -c conda-forge -y \
     samtools \
     htslib \
     pysam \ 
-    cooler && conda clean -afy
+    cooler \
+    pairtools && conda clean -afy
 
 RUN pip install -Ur requirements.txt
 # Using pip:

environment.yml

@@ -1,4 +1,4 @@
-name: env-name
+name: base
 channels:
   - conda-forge
   - bioconda
@@ -21,4 +21,5 @@ dependencies:
   - pysam
   - htslib
   - pyfastx
-  - cooler
+  - cooler
+  - pairtools

hicstuff/pipeline.py

@@ -1101,9 +1101,17 @@ def _out_file(fname):
 
     # Move final pairs file to main dir. 
     p = pathlib.Path(use_pairs).absolute()
-    parent_dir = p.parents[1]
-    p.rename(parent_dir / p.name)
-
+    pairsf = p.parents[1] / p.name
+    p.rename(pairsf)
+
+    # Sort and compress final pairs file
+    pairstools_cmd = "pairtools sort".split(" ")
+    sorted_pairsf = str(pairsf) + ".gz"
+    sort_args = "--output {out} --tmpdir {tmp_dir}".format(
+        out = sorted_pairsf, tmp_dir = tmp_dir
+    )
+    sp.call(pairstools_cmd + sort_args.split(" ") + [pairsf], shell=False)
+
     # Clean temporary files
     if not no_cleanup:
         tempfiles = [

requirements.txt

@@ -9,3 +9,4 @@ scikit-learn
 pysam
 pyfastx
 cooler
+pairtools