flexible disk space allocation for RNA-seq

Author: clarabakker

pyproject.toml

@@ -4,7 +4,7 @@
 
 [tool.poetry]
 name = "Benchmark-4dn"
-version = "0.5.25"
+version = "0.5.26"
 description = """Benchmark functions that returns total space, mem, cpu given \
     input size and parameters for the CWL workflows"""
 authors = ["Soo Lee <[email protected]>"]

src/Benchmark/bfunctions.py

@@ -4,7 +4,19 @@
 
 def encode_rnaseq_stranded(input_json):
     assert 'input_size_in_bytes' in input_json
-    r = BenchmarkResult(size=300,
+    insz = input_json['input_size_in_bytes']
+    input_sizes = 0
+    for file,file_size in insz.items():
+        input_sizes += file_size
+    # The starIndex file is the main input size difference, meaning
+    # disk requirements vary by organism. 24 GB+ is usually mouse and
+    # < 24 GB is usually human. As human sets approach 24 GB, the disk
+    # requirement is less predictable, so disk may need to manually set.
+    if B2GB(input_sizes) < 24:
+        total_size_in_gb = B2GB((input_sizes) * 7) + 21
+    else:
+        total_size_in_gb = B2GB((input_sizes) * 10.1) - 150
+    r = BenchmarkResult(size=total_size_in_gb,
                         mem=GB2MB(64),
                         cpu=16,
                         exclude_t=True)
@@ -13,7 +25,21 @@ def encode_rnaseq_stranded(input_json):
 
 def encode_rnaseq_unstranded(input_json):
     assert 'input_size_in_bytes' in input_json
-    r = BenchmarkResult(size=300,
+    insz = input_json['input_size_in_bytes']
+    assert 'rna.align_index' in insz
+    input_sizes = 0
+    for file,file_size in insz.items():
+        # As in stranded case, align index is the main contributor,
+        # but now total file size cannot determine organism.
+        # Instead, use the size of the align index itself
+        if file == 'rna.align_index':
+            mouse = B2GB(file_size) > 13
+        input_sizes += file_size
+    if mouse:
+        total_size_in_gb = B2GB((input_sizes) * 8) - 105
+    else:
+        total_size_in_gb = B2GB((input_sizes) * 5.5) + 35
+    r = BenchmarkResult(size=total_size_in_gb,
                         mem=GB2MB(64),
                         cpu=16,
                         exclude_t=True)

tests/test.py

@@ -21,23 +21,27 @@ class TestBenchmark(unittest.TestCase):
     def test_encode_rnaseq_unstranded(self):
         print("rnaseq_unstranded")
         input_json = {'input_size_in_bytes': {'rna.fastqs_R1': GB2B(10),
-                                              'rna.align_index': GB2B(3)}}
+                                              'rna.align_index': GB2B(10)}}
         res = B.benchmark('encode-rnaseq-unstranded', input_json)
         print(res)
         assert 'aws' in res
         assert 'recommended_instance_type' in res['aws']
         assert res['aws']['recommended_instance_type'] == 'm5a.4xlarge'
+        assert 'total_size_in_GB' in res
+        assert int(res['total_size_in_GB']) == 145
 
 
     def test_encode_rnaseq_stranded(self):
         print("rnaseq_stranded")
         input_json = {'input_size_in_bytes': {'rna.fastqs_R1': GB2B(10),
-                                              'rna.align_index': GB2B(3)}}
+                                              'rna.align_index': GB2B(10)}}
         res = B.benchmark('encode-rnaseq-stranded', input_json)
         print(res)
         assert 'aws' in res
         assert 'recommended_instance_type' in res['aws']
         assert res['aws']['recommended_instance_type'] == 'm5a.4xlarge'
+        assert 'total_size_in_GB' in res
+        assert int(res['total_size_in_GB']) == 161
 
 
     def test_repliseq(self):