write cutoff network files and full network file

medema-group · Nov 3, 2023 · 4b0693c · 4b0693c
1 parent e2f4751
commit 4b0693c
Show file tree

Hide file tree

Showing 3 changed files with 286 additions and 25 deletions.
diff --git a/big_scape/main.py b/big_scape/main.py
@@ -23,7 +23,7 @@
     legacy_prepare_bin_output,
     legacy_generate_bin_output,
     write_record_annotations_file,
-    write_clustering_file,
+    write_full_network_file,
 )
 
 
@@ -341,6 +341,9 @@ def align_callback(tasks_done: int):
     # prepare output files
     legacy_prepare_output(run["output_dir"], pfam_info)
 
+    # write full network file
+    write_full_network_file(run, all_bgc_records)
+
     # prepare output files per cutoff
     for cutoff in run["gcf_cutoffs"]:
         # TODO: update to use records and not gbk regions?
@@ -364,9 +367,8 @@ def align_callback(tasks_done: int):
                         f"Network {mix_bin.label} with cutoff {cutoff} is empty after culling singletons"
                     )
                     continue
-            legacy_prepare_bin_output(run["output_dir"], run["label"], cutoff, mix_bin)
-            legacy_generate_bin_output(run["output_dir"], run["label"], cutoff, mix_bin)
-            write_clustering_file(run, cutoff, mix_bin)
+            legacy_prepare_bin_output(run, cutoff, mix_bin)
+            legacy_generate_bin_output(run, cutoff, mix_bin)
 
     # legacy_classify
 
@@ -382,9 +384,8 @@ def align_callback(tasks_done: int):
                             f"Network '{bin.label}' with cutoff {cutoff} is empty after culling singletons"
                         )
                         continue
-                legacy_prepare_bin_output(run["output_dir"], run["label"], cutoff, bin)
-                legacy_generate_bin_output(run["output_dir"], run["label"], cutoff, bin)
-                write_clustering_file(run, cutoff, bin)
+                legacy_prepare_bin_output(run, cutoff, bin)
+                legacy_generate_bin_output(run, cutoff, bin)
 
     # classify
 
@@ -409,9 +410,8 @@ def align_callback(tasks_done: int):
                             f"Network '{bin.label}' with cutoff {cutoff} is empty after culling singletons"
                         )
                         continue
-                legacy_prepare_bin_output(run["output_dir"], run["label"], cutoff, bin)
-                legacy_generate_bin_output(run["output_dir"], run["label"], cutoff, bin)
-                write_clustering_file(run, cutoff, bin)
+                legacy_prepare_bin_output(run, cutoff, bin)
+                legacy_generate_bin_output(run, cutoff, bin)
 
     # query
 
@@ -423,13 +423,8 @@ def align_callback(tasks_done: int):
             [record for record in all_bgc_records if record is not None]
         )
         for cutoff in run["gcf_cutoffs"]:
-            legacy_prepare_bin_output(
-                run["output_dir"], run["label"], cutoff, query_bin
-            )
-            legacy_generate_bin_output(
-                run["output_dir"], run["label"], cutoff, query_bin
-            )
-            write_clustering_file(run, cutoff, query_bin)
+            legacy_prepare_bin_output(run, cutoff, query_bin)
+            legacy_generate_bin_output(run, cutoff, query_bin)
 
     if run["profiling"]:
         profiler.stop()

diff --git a/big_scape/output/__init__.py b/big_scape/output/__init__.py
@@ -5,6 +5,8 @@
     legacy_generate_bin_output,
     write_record_annotations_file,
     write_clustering_file,
+    write_cutoff_network_file,
+    write_full_network_file,
 )
 
 __all__ = [
@@ -14,4 +16,6 @@
     "legacy_generate_bin_output",
     "write_record_annotations_file",
     "write_clustering_file",
+    "write_cutoff_network_file",
+    "write_full_network_file",
 ]