Recyze updates (#523)

* Apply proper ome-xml encoding

This avoids errors from tifffile about the imagedescription value, allowing us
to use newer versions of ome-types which are faster and produce nicer output.

* Add parallelization management

* Improve --channels help message

* Bump default roadie version and switch to ghcr

nextflow.config

@@ -1,5 +1,5 @@
 // Execution environment for miscellaneous tasks
-params.roadie = 'labsyspharm/roadie:2023-03-08'
+params.roadie = 'ghcr.io/labsyspharm/mcmicro:roadie-2023-10-25'
 
 // Platform-specific profiles
 profiles {

roadie/scripts/recyze.py

@@ -223,7 +223,7 @@ def run(self):
                 self.metadata.images[0].pixels.tiff_data_blocks[0].plane_count = self.num_channels
 
             # Write
-        tifffile.tiffcomment(self.out_path, to_xml(self.metadata))
+        tifffile.tiffcomment(self.out_path, to_xml(self.metadata).encode())
 
 
 if __name__ == '__main__':
@@ -236,7 +236,14 @@ def run(self):
     parser.add_argument('--y2', type=int, required=False, default=None, help="Crop Y2")
     parser.add_argument('--w', type=int, required=False, default=None, help="Crop Width")
     parser.add_argument('--h', type=int, required=False, default=None, help="Crop Height")
-    parser.add_argument('--channels', type=int, nargs="+", required=False, default=None, help="Channels")
+    parser.add_argument(
+        '--channels', type=int, nargs="+", required=False, default=None, metavar="C",
+        help="Channels to keep (Default: all)",
+    )
+    parser.add_argument(
+        '--num-threads', type=int, required=False, default=0, metavar="N",
+        help="Worker thread count (Default: auto-scale based on number of available CPUs)",
+    )
     argument = parser.parse_args()
 
     # Automatically infer the output filename, if not specified
@@ -256,6 +263,15 @@ def run(self):
             stem = os.extsep.join(tokens[0:-1]) + "_crop"
             out_path = os.extsep.join([stem, tokens[-1]])
 
+    num_threads = argument.num_threads
+    if num_threads == 0:
+        if hasattr(os, "sched_getaffinity"):
+            num_threads = len(os.sched_getaffinity(0))
+        else:
+            num_threads = os.cpu_count()
+    tifffile.TIFF.MAXWORKERS = num_threads
+    tifffile.TIFF.MAXIOWORKERS = num_threads * 5
+
     writer = PyramidWriter(in_path, out_path, argument.channels, argument.x, argument.y,
                            argument.x2, argument.y2, argument.w, argument.h)
     writer.run()