add --bridge-gaps option to diff

methscan/cli.py

@@ -403,11 +403,11 @@ def smooth_cli(**kwargs):
 )
 @click.option(
     "--bridge-gaps",
-    metavar="INTEGER",
     default=0,
-    type=int,
+    type=click.IntRange(min=0),
+    metavar="INTEGER",
     help="Merge neighboring VMRs if they are within this distance in basepairs. Useful "
-    "to prevent fragmented VMRs separated only by small gaps.  [default: off; x>=0]",
+    "to prevent fragmented VMRs separated only by small gaps.  [default: off]",
 )
 @click.option(
     "--threads",
@@ -503,6 +503,14 @@ def scan_cli(**kwargs):
     "testing. For example, a value of 6 means that only regions with sequencing "
     "coverage in at least 6 cells per group are considered.",
 )
+@click.option(
+    "--bridge-gaps",
+    default=0,
+    type=click.IntRange(min=0),
+    metavar="INTEGER",
+    help="Merge neighboring VMRs if they are within this distance in basepairs. Useful "
+    "to prevent fragmented VMRs separated only by small gaps.  [default: off]",
+)
 @click.option(
     "--threads",
     default=-1,
@@ -637,14 +645,14 @@ def matrix_cli(**kwargs):
     type=click.IntRange(min=1),
     metavar="INTEGER",
     help="The bed column number (1-indexed) denoting "
-    "the DNA strand of the region  [optional].",
+    "the DNA strand of the region.  [optional]",
 )
 @click.option(
     "--label",
     help="Specify a constant value to be added as a "
     "column to the output table. This can be "
     "useful to give each output a unique label when "
-    "you want to concatenate multiple outputs  [optional].",
+    "you want to concatenate multiple outputs.  [optional]",
 )
 def profile_cli(**kwargs):
     from .profile import profile

methscan/diff.py

@@ -238,6 +238,7 @@ def calc_tstat_peaks(
     chrom_len,
     index,
     min_cells,
+    bridge_gaps,
     threshold_datatype,
     window_tstat_groups,
     genomic_positions,
@@ -252,7 +253,7 @@ def calc_tstat_peaks(
         # merge overlapping windows with lowest and highest t-statistic,
         # to get bigger regions of variable size
         peak_starts, peak_ends = _find_peaks(
-            tstat_windows, genomic_positions, threshold_value, half_bw
+            tstat_windows, genomic_positions, threshold_value, half_bw, bridge_gaps
         )
 
         # for each big merged peak, re-calculate the t-statistic
@@ -389,6 +390,7 @@ def diff(
     stepsize,
     threshold,
     min_cells,
+    bridge_gaps,
     threads=-1,
     write_header=False,
     debug=False,
@@ -526,6 +528,7 @@ def diff(
                 chrom_len,
                 perm_idx,
                 min_cells,
+                bridge_gaps,
                 threshold_datatype,
                 window_tstat_groups,
                 genomic_positions,