Not require sorted bed

pygenometracks/readBed.py

@@ -38,9 +38,6 @@ def __init__(self, file_handle):
         else:
             self.get_bed_interval(fields, is_first_line=True)
         self.file_handle.seek(0)
-        self.prev_chrom = None
-        self.prev_start = -1
-        self.prev_line = None
 
         # list of bed fields
         self.fields = ['chromosome', 'start', 'end',
@@ -81,15 +78,6 @@ def next(self):
         line = self.get_no_comment_line()
 
         bed = self.get_bed_interval(line)
-        if self.prev_chrom == bed.chromosome:
-            assert self.prev_start <= bed.start, \
-                "Bed file not sorted. Please use a sorted bed file.\n" \
-                f"File: {self.file_handle.name}\n" \
-                f"Previous line: {self.prev_line}\n Current line{line} "
-
-        self.prev_chrom = bed.chromosome
-        self.prev_start = bed.start
-        self.prev_line = line
 
         return bed
 
@@ -100,15 +88,6 @@ def __next__(self):
         line = self.get_no_comment_line()
 
         bed = self.get_bed_interval(line)
-        if self.prev_chrom == bed.chromosome:
-            assert self.prev_start <= bed.start, \
-                "Bed file not sorted. Please use a sorted bed file.\n" \
-                f"File: {self.file_handle.name}\n" \
-                f"Previous line: {self.prev_line}\n Current line{line}"
-
-        self.prev_chrom = bed.chromosome
-        self.prev_start = bed.start
-        self.prev_line = line
 
         return bed
 

pygenometracks/tests/generateAllOutput.sh

@@ -13,6 +13,7 @@ bin/pgt --tracks ./pygenometracks/tests/test_data/bed_all_labels_inside.ini --re
 bin/pgt --tracks ./pygenometracks/tests/test_data/bed_all_labels_inside.ini --region X:3215000-3240000 --trackLabelFraction 0.2 --width 38 --dpi 130 --trackLabelHAlign right --decreasingXAxis -o ./pygenometracks/tests/test_data/master_bed_all_label_inside_dec.png
 bin/pgt --tracks pygenometracks/tests/test_data/bed_unusual_formats.ini --region X:20000-40000 --trackLabelFraction 0.2 --width 38 --dpi 130 -o pygenometracks/tests/test_data/master_bed_unusual_formats.png
 bin/pgt --tracks ./pygenometracks/tests/test_data/bed_colormap_genes.ini --region X:3000000-3300000 --trackLabelFraction 0.2 --width 38 --dpi 130 -o ./pygenometracks/tests/test_data/master_bed_colormap_genes.png
+bin/pgt --tracks ./pygenometracks/tests/test_data/bed_shuffle.ini --BED ./pygenometracks/tests/test_data/regions_chr1XY.bed --trackLabelFraction 0.2 --width 38 --dpi 130 -o ./pygenometracks/tests/test_data/master_bed_shuffle.png
 
 # test_bedGraphMatrixTrack:
 bin/pgt --tracks ./pygenometracks/tests/test_data/bedgraph.ini --region X:2850000-3150000 --trackLabelFraction 0.2 --dpi 130 -o ./pygenometracks/tests/test_data/master_bedgraph.png

pygenometracks/tests/test_bed_and_gtf_tracks.py

@@ -549,6 +549,35 @@
 with open(os.path.join(ROOT, "bed_colormap_genes.ini"), 'w') as fh:
     fh.write(browser_tracks)
 
+browser_tracks = """
+[genes]
+file = dm3_genes_withrgbandscore.bed.gz
+title = bed color = Reds
+color = Reds
+height = 4
+
+[spacer]
+
+[genes]
+file = dm3_genes_withrgbandscore_shuffled.bed.gz
+title = bed color = Reds bed is not sorted
+color = Reds
+height = 4
+
+[spacer]
+
+[genes]
+file = dm3_genes_withrgbandscore_shuffled.bed.gz
+title = bed color = Reds global_max_row = true and bed is not sorted
+color = Reds
+global_max_row = true
+height = 4
+
+[x-axis]
+"""
+with open(os.path.join(ROOT, "bed_shuffle.ini"), 'w') as fh:
+    fh.write(browser_tracks)
+
 tolerance = 13  # default matplotlib pixed difference tolerance
 
 
@@ -822,3 +851,26 @@ def test_plot_tracks_bed_scores():
     assert res is None, res
 
     os.remove(outfile.name)
+
+
+def test_bed_shuffle():
+    extension = '.png'
+
+    outfile = NamedTemporaryFile(suffix=extension, prefix='pyGenomeTracks_test_',
+                                 delete=False)
+    ini_file = os.path.join(ROOT, "bed_shuffle.ini")
+    bed_file = os.path.join(ROOT, 'regions_chr1XY.bed')
+    args = f"--tracks {ini_file} --BED {bed_file} "\
+           "--trackLabelFraction 0.2 --width 38 --dpi 130 "\
+           f"--outFileName {outfile.name}".split()
+    pygenometracks.plotTracks.main(args)
+    for region in ['chr1:0-500000', 'chrX:2500000-2600000', 'chrY:0-1000000']:
+        region_str = region.replace(':', '-')
+        output_file = outfile.name[:-4] + '_' + region_str + extension
+        expected_file = os.path.join(ROOT, 'master_bed_shuffle_'
+                                     + region_str + extension)
+        res = compare_images(expected_file,
+                             output_file, tolerance)
+        assert res is None, res
+
+        os.remove(output_file)

pygenometracks/tests/test_data/bed_shuffle.ini

@@ -0,0 +1,25 @@
+
+[genes]
+file = dm3_genes_withrgbandscore.bed.gz
+title = bed color = Reds
+color = Reds
+height = 4
+
+[spacer]
+
+[genes]
+file = dm3_genes_withrgbandscore_shuffled.bed.gz
+title = bed color = Reds bed is not sorted
+color = Reds
+height = 4
+
+[spacer]
+
+[genes]
+file = dm3_genes_withrgbandscore_shuffled.bed.gz
+title = bed color = Reds global_max_row = true and bed is not sorted
+color = Reds
+global_max_row = true
+height = 4
+
+[x-axis]

pygenometracks/tests/test_data/browser_tracks_hic_small_test.ini

@@ -16,4 +16,8 @@ file_type = hic_matrix
 height = 5
 show_masked_bins = true
 
+[spacer]
+
+[genes]
+file = dm3_genes_withrgbandscore_shuffled.bed.gz
 [x-axis]

pygenometracks/utilities.py

@@ -105,9 +105,6 @@ def file_to_intervaltree(file_name, plot_regions=None):
     file_h = opener(file_to_open)
     line_number = 0
     valid_intervals = 0
-    prev_chrom = None
-    prev_start = -1
-    prev_line = None
     interval_tree = {}
     min_value = float('Inf')
     max_value = -float('Inf')
@@ -138,10 +135,6 @@ def file_to_intervaltree(file_name, plot_regions=None):
                   f"an integer.\nError message: {detail}"
             raise InputError(msg)
 
-        if prev_chrom == chrom:
-            assert prev_start <= start, \
-                f"Bed file not sorted. Please use a sorted bed file.\n{prev_line}{line} "
-
         if chrom not in interval_tree:
             interval_tree[chrom] = IntervalTree()