Fix mosdepth coverage in general stats if key is missing (MultiQC#3432)

Author: jchorl

docs/markdown/modules/mosdepth.md

@@ -54,6 +54,17 @@ mosdepth_config:
     - 30000
 ```
 
+:::info{title="Missing coverage thresholds"}
+
+Mosdepth does omit cumulative coverages at high coverage thresholds. To work around this, MultiQC will
+use the next cumulative coverage available. E.g. if 3301x coverage is present, but 3300x is missing,
+and 3300x is requested, the value of 3301x will be used. This should provide the technically correct value
+because it is a cumulative distribution and mosdepth [only skips values if there are no bases at that coverage
+level](https://github.com/brentp/mosdepth/blob/df8c74a0aabae4ae5e6da1ea0c633b4db829f20a/mosdepth.nim#L439).
+
+For more details, see [this](https://github.com/MultiQC/MultiQC/pull/3432#issuecomment-3672811525) comment.
+:::
+
 You can also specify which columns would be hidden when the report loads (by default, all values are hidden except 30X):
 
 ```yaml
@@ -125,4 +136,4 @@ mosdepth/region_dist:
 mosdepth/summary:
   fn: '*.mosdepth.summary.txt'
 ```
-    
+    

multiqc/modules/mosdepth/mosdepth.py

@@ -57,8 +57,13 @@ def read_config():
 
 def genstats_cov_thresholds(cum_fraction_by_cov: Dict[int, float], threshs: List[int]) -> Dict[str, float]:
     genstats: Dict[str, float] = {}
+    sorted_cum_fraction_by_cov = sorted(cum_fraction_by_cov.items())
     for t in threshs:
-        genstats[f"{t}_x_pc"] = cum_fraction_by_cov.get(t, 0.0) * 100.0
+        # take next known value
+        # e.g. if only 50x is known but the threshold is 40x, take the 50x value
+        # if there is no next threshold, just take 0.0
+        cov_val = next((proportion for cov, proportion in sorted_cum_fraction_by_cov if cov >= t), 0.0)
+        genstats[f"{t}_x_pc"] = cov_val * 100.0
     return genstats
 
 

multiqc/modules/mosdepth/tests/__init__.py

@@ -0,0 +1,21 @@
+import pytest
+
+from multiqc.modules.mosdepth.mosdepth import genstats_cov_thresholds
+
+
+def test_genstats_cov_thresholds():
+    cum_fraction_by_cov = {
+        1: 1.0,
+        10: 0.8,
+        20: 0.2,
+        30: 0.1,
+    }
+    thresholds = 10, 15, 30, 200
+
+    actual_thresholds = genstats_cov_thresholds(cum_fraction_by_cov, thresholds)
+    assert actual_thresholds == {
+        "10_x_pc": 80.0,
+        "15_x_pc": 20.0,
+        "30_x_pc": 10.0,
+        "200_x_pc": 0.0,
+    }