From 51f1077ca8b5f9d05ce85538e8d511b51b2a4ffc Mon Sep 17 00:00:00 2001
From: _ <4256466+leehart@users.noreply.github.com>
Date: Fri, 13 Sep 2024 11:10:57 +0100
Subject: [PATCH] Skip sets with missing CNV DRCs rather than
 FileNotFoundError. Raise ValueError when no given sets have DRCs.

---
 malariagen_data/anoph/cnv_data.py | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

diff --git a/malariagen_data/anoph/cnv_data.py b/malariagen_data/anoph/cnv_data.py
index 38e389939..f9e6d474c 100644
--- a/malariagen_data/anoph/cnv_data.py
+++ b/malariagen_data/anoph/cnv_data.py
@@ -454,7 +454,11 @@ def _cnv_discordant_read_calls_dataset(
         data_vars = dict()
 
         debug("open zarr")
-        root = self.open_cnv_discordant_read_calls(sample_set=sample_set)
+
+        try:
+            root = self.open_cnv_discordant_read_calls(sample_set=sample_set)
+        except FileNotFoundError:
+            return None
 
         # not all contigs have CNVs, need to check
         # TODO consider returning dataset with zero length variants dimension, would
@@ -556,8 +560,17 @@ def cnv_discordant_read_calls(
                     inline_array=inline_array,
                     chunks=chunks,
                 )
+
+                # If no CNV DRCs dataset was found then skip
+                if y is None:
+                    continue
+
                 ly.append(y)
 
+                if len(ly) == 0:
+                    # Bail out, no data for given sample sets and analysis.
+                    raise ValueError("No data found for requested sample sets.")
+
             x = simple_xarray_concat(ly, dim=DIM_SAMPLE)
             lx.append(x)