Add initial qptiff cleaning

bin/clean_qptiff.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python
+
+import sys
+import tifftools
+import re
+import os
+
+input = sys.argv[1]
+
+os.rename(input, input + ".original")
+
+date_pattern = r"(<Date>((\d{4}-\d{2}-\d{2})T(\d{2}:\d{2}:\d{2})(\.\d+)?Z?))</Date>"
+date_replacement = "<Date />"
+
+
+sample_pattern = r"<SampleDescription>.+</SampleDescription>"
+sample_replacement = "<SampleDescription />"
+
+qptiff_info = tifftools.read_tiff(input + ".original")
+
+for i, ifd in enumerate(qptiff_info["ifds"]):
+    for tag in ifd["tags"]:
+        if tag == tifftools.Tag.IMAGEDESCRIPTION:
+            old_description = ifd["tags"][tifftools.Tag.IMAGEDESCRIPTION.value]["data"]
+            new_description = re.sub(date_pattern, date_replacement, old_description)
+            new_description = re.sub(sample_pattern, sample_replacement, new_description)
+            if new_description != old_description:
+                print(f"Redacting date in IFD {i}, tag {tag}")
+                qptiff_info["ifds"][i]["tags"][tifftools.Tag.IMAGEDESCRIPTION.value][
+                    "data"
+                ] = new_description
+
+tifftools.write_tiff(qptiff_info, input, allowExisting=True)

bin/clean_tiff.py

@@ -29,7 +29,7 @@ def split_all_ext(filename):
 print(new_filename)
 
 
-unset_list = [
+base_unset_list = [
     "DateTime",
     "NDPI_ScanTime",
     "NDPI_WriteTime",
@@ -46,9 +46,18 @@ def split_all_ext(filename):
     "45494"
 ]
 
+info = tifftools.read_tiff(args.input)
+num_ifds = len(info['ifds'])
+
+unset_list_for_all_ifds = []
+for ifd in range(num_ifds):
+    unset_list_for_ifd = [f"{tag},{ifd}" for tag in base_unset_list]
+    unset_list_for_all_ifds.extend(unset_list_for_ifd)
+
+# Call tiff_set only once after all IFD modifications are specified
 tifftools.tiff_set(
     args.input,
     output=new_filename,
     overwrite=False,
-    unset=unset_list
-)
+    unset=unset_list_for_all_ifds
+)

modules/clean_qptiff.nf

@@ -0,0 +1,14 @@
+process clean_qptiff {
+  input:
+    tuple val(meta), file(image)
+  output:
+    tuple val(meta), file(image)
+  stub:
+  """
+  touch image_cleaned.svs
+  """
+  script:
+  """
+  clean_qptiff.py $image
+  """
+}

subworkflows/clean_qptiff.nf

@@ -0,0 +1,17 @@
+//include { remove_qptiff_label } from "../modules/remove_qptiff_label.nf"
+//include { remove_qptiff_macro } from "../modules/remove_qptiff_macro.nf"
+include { clean_qptiff } from "../modules/clean_qptiff.nf"
+
+
+workflow CLEAN_QPTIFF {
+  take:
+  images
+
+  main:
+
+    clean_qptiff(images)
+    clean_qptiff.out.set{cleaned}
+
+  emit:
+  cleaned
+}

subworkflows/samplesheet_split.nf

@@ -41,12 +41,14 @@ workflow SAMPLESHEET_SPLIT {
             .branch {
                 ome: it[1] =~ /.+\.ome\.tif{1,2}$/
                 svs: it[1] =~ /.+\.svs$/
+                qptiff: it[1] =~ /.+\.qptiff$/
                 other: true
             }
             .set{ images }
 
     emit:
         ome = images.ome
         svs = images.svs
+        qptiff = images.qptiff
         other = images.other
 }

workflows/cleaner.nf

@@ -1,14 +1,16 @@
 include { SAMPLESHEET_SPLIT } from '../subworkflows/samplesheet_split.nf'
 include { CLEAN_OME } from '../subworkflows/clean_ome.nf'
 include { CLEAN_SVS } from '../subworkflows/clean_svs.nf'
+include { CLEAN_QPTIFF } from '../subworkflows/clean_qptiff.nf'
 include { CLEAN_TIFF } from '../subworkflows/clean_tiff.nf'
 
 workflow CLEANER {
     SAMPLESHEET_SPLIT (params.input)
-    CLEAN_OME ( SAMPLESHEET_SPLIT.out.ome )
-    CLEAN_SVS ( SAMPLESHEET_SPLIT.out.svs )
+    CLEAN_OME    ( SAMPLESHEET_SPLIT.out.ome )
+    CLEAN_SVS    ( SAMPLESHEET_SPLIT.out.svs )
+    CLEAN_QPTIFF ( SAMPLESHEET_SPLIT.out.qptiff )
     SAMPLESHEET_SPLIT.out.other
-        .mix ( CLEAN_OME.out.cleaned, CLEAN_SVS.out.cleaned )
+        .mix ( CLEAN_OME.out.cleaned, CLEAN_SVS.out.cleaned, CLEAN_QPTIFF.out.cleaned )
         .set { combined }
     CLEAN_TIFF ( combined )
 }