From 3fe893328d7590364f7f47e334008b2050edde0e Mon Sep 17 00:00:00 2001
From: martinghunt <martin.g.hunt@gmail.com>
Date: Thu, 28 Apr 2016 14:40:32 +0000
Subject: [PATCH 1/2] Add method _filter_clusters

---
 ariba/summary.py            | 30 ++++++++++++++++++++++++++++++
 ariba/tests/summary_test.py | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 64 insertions(+)

diff --git a/ariba/summary.py b/ariba/summary.py
index 82fab952..881ac7b3 100644
--- a/ariba/summary.py
+++ b/ariba/summary.py
@@ -114,6 +114,36 @@ def _gather_output_rows(self):
         return rows
 
 
+    @classmethod
+    def _filter_clusters(cls, rows):
+        '''Removes any cluster where every sample has assembled == "no"'''
+        found_a_yes = set()
+        first_filename = True
+        all_clusters = set()
+
+        for filename in rows:
+            for cluster in rows[filename]:
+                if first_filename:
+                    all_clusters.add(cluster)
+
+                if cluster in found_a_yes:
+                    continue
+
+                if rows[filename][cluster]['assembled'] == 'yes':
+                    found_a_yes.add(cluster)
+                    break
+
+            first_filename = False
+
+        to_delete = all_clusters.difference(found_a_yes)
+
+        for filename in rows:
+            for cluster in to_delete:
+                del rows[filename][cluster]
+
+        return rows
+
+
     @classmethod
     def _write_csv(cls, filenames, rows, outfile, phandango=False):
         lines = []
diff --git a/ariba/tests/summary_test.py b/ariba/tests/summary_test.py
index 0b7a9716..fc13b01c 100644
--- a/ariba/tests/summary_test.py
+++ b/ariba/tests/summary_test.py
@@ -120,6 +120,40 @@ def test_gather_output_rows(self):
         self.assertEqual(expected, got)
 
 
+    def test_filter_clusters(self):
+        '''Test _filter_clusters'''
+        rows = {
+            'file1': {
+                'cluster1': {'assembled': 'yes'},
+                'cluster2': {'assembled': 'yes'},
+                'cluster3': {'assembled': 'no'},
+                'cluster4': {'assembled': 'no'},
+            },
+            'file2': {
+                'cluster1': {'assembled': 'yes'},
+                'cluster2': {'assembled': 'no'},
+                'cluster3': {'assembled': 'yes'},
+                'cluster4': {'assembled': 'no'},
+            }
+        }
+
+        expected = {
+            'file1': {
+                'cluster1': {'assembled': 'yes'},
+                'cluster2': {'assembled': 'yes'},
+                'cluster3': {'assembled': 'no'},
+            },
+            'file2': {
+                'cluster1': {'assembled': 'yes'},
+                'cluster2': {'assembled': 'no'},
+                'cluster3': {'assembled': 'yes'},
+            }
+        }
+
+        got = summary.Summary._filter_clusters(rows)
+        self.assertEqual(expected, got)
+
+
     def test_write_csv(self):
         '''Test _write_csv'''
         tmp_out = 'tmp.out.tsv'

From 4f0caffcfee41d133b24f080461c2fb03e5d9c66 Mon Sep 17 00:00:00 2001
From: martinghunt <martin.g.hunt@gmail.com>
Date: Thu, 28 Apr 2016 14:53:00 +0000
Subject: [PATCH 2/2] Report error if all columns get removed

---
 ariba/summary.py            | 11 ++++++++---
 ariba/tests/summary_test.py |  2 +-
 2 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/ariba/summary.py b/ariba/summary.py
index 881ac7b3..3e354f68 100644
--- a/ariba/summary.py
+++ b/ariba/summary.py
@@ -116,7 +116,8 @@ def _gather_output_rows(self):
 
     @classmethod
     def _filter_clusters(cls, rows):
-        '''Removes any cluster where every sample has assembled == "no"'''
+        '''Removes any cluster where every sample has assembled == "no".
+           Returns tuple: (filtered rows, number of remaining columns)'''
         found_a_yes = set()
         first_filename = True
         all_clusters = set()
@@ -131,7 +132,6 @@ def _filter_clusters(cls, rows):
 
                 if rows[filename][cluster]['assembled'] == 'yes':
                     found_a_yes.add(cluster)
-                    break
 
             first_filename = False
 
@@ -141,7 +141,7 @@ def _filter_clusters(cls, rows):
             for cluster in to_delete:
                 del rows[filename][cluster]
 
-        return rows
+        return rows, len(found_a_yes)
 
 
     @classmethod
@@ -253,6 +253,11 @@ def run(self):
         self._check_files_exist()
         self.samples = self._load_input_files(self.filenames, self.min_id)
         self.rows = self._gather_output_rows()
+        self.rows, remaining_clusters = Summary._filter_clusters(self.rows)
+        if remaining_clusters == 0:
+            print('No clusters found that are present in any sample. Will not write any output files', file=sys.stderr)
+            sys.exit(1)
+
         Summary._write_csv(self.filenames, self.rows, self.outprefix + '.csv', phandango=False)
 
         if len(self.samples) > 1:
diff --git a/ariba/tests/summary_test.py b/ariba/tests/summary_test.py
index fc13b01c..8f5094e3 100644
--- a/ariba/tests/summary_test.py
+++ b/ariba/tests/summary_test.py
@@ -151,7 +151,7 @@ def test_filter_clusters(self):
         }
 
         got = summary.Summary._filter_clusters(rows)
-        self.assertEqual(expected, got)
+        self.assertEqual((expected, 3), got)
 
 
     def test_write_csv(self):