Merge pull request #25 from andersen-lab/dev

adding custom lineage metadata file for reproducibility

freyja/_cli.py

@@ -25,9 +25,10 @@ def cli():
 @click.argument('depths', type=click.Path(exists=True))
 @click.option('--eps', default=1e-3, help='minimum abundance to include')
 @click.option('--barcodes', default='-1', help='custom barcode file')
+@click.option('--meta', default='-1', help='custom lineage metadata file')
 @click.option('--output', default='demixing_result.csv', help='Output file',
               type=click.Path(exists=False))
-def demix(variants, depths, output, eps, barcodes):
+def demix(variants, depths, output, eps, barcodes, meta):
     locDir = os.path.abspath(os.path.join(os.path.realpath(__file__),
                              os.pardir))
     # option for custom barcodes
@@ -37,7 +38,7 @@ def demix(variants, depths, output, eps, barcodes):
         df_barcodes = pd.read_csv(os.path.join(locDir,
                                   'data/usher_barcodes.csv'), index_col=0)
     muts = list(df_barcodes.columns)
-    mapDict = buildLineageMap()
+    mapDict = buildLineageMap(meta)
     print('building mix/depth matrices')
     # assemble data from (possibly) mixed samples
     if variants.lower().endswith('vcf'):
@@ -115,9 +116,10 @@ def variants(bamfile, ref, variants, depths, refname):
 @click.option('--nt', default=1, help='max number of cpus to use')
 @click.option('--eps', default=1e-3, help='minimum abundance to include')
 @click.option('--barcodes', default='-1', help='custom barcode file')
+@click.option('--meta', default='-1', help='custom lineage metadata file')
 @click.option('--output_base', default='test', help='Output file basename',
               type=click.Path(exists=False))
-def boot(variants, depths, output_base, eps, barcodes, nb, nt):
+def boot(variants, depths, output_base, eps, barcodes, meta, nb, nt):
     locDir = os.path.abspath(os.path.join(os.path.realpath(__file__),
                              os.pardir))
     # option for custom barcodes
@@ -127,7 +129,7 @@ def boot(variants, depths, output_base, eps, barcodes, nb, nt):
         df_barcodes = pd.read_csv(os.path.join(locDir,
                                   'data/usher_barcodes.csv'), index_col=0)
     muts = list(df_barcodes.columns)
-    mapDict = buildLineageMap()
+    mapDict = buildLineageMap(meta)
     print('building mix/depth matrices')
     # assemble data from (possibly) mixed samples
     mix, depths_ = build_mix_and_depth_arrays(variants, depths, muts)

freyja/sample_deconv.py

@@ -9,13 +9,18 @@
 from tqdm import tqdm
 
 
-def buildLineageMap():
+def buildLineageMap(locDir):
     # Parsing curated lineage data from outbreak.info
-    locDir = os.path.abspath(os.path.join(os.path.realpath(__file__),
-                             os.pardir))
-    f0 = open(os.path.join(locDir, 'data/curated_lineages.json'))
-    dat = json.load(f0)
-    f0.close()
+    if locDir == '-1':
+        locDir = os.path.abspath(os.path.join(os.path.realpath(__file__),
+                                 os.pardir))
+        f0 = open(os.path.join(locDir, 'data/curated_lineages.json'))
+        dat = json.load(f0)
+        f0.close()
+    else:
+        f0 = open(locDir)
+        dat = json.load(f0)
+        f0.close()
 
     mapDict = {}
     for ind in range(len(dat)):

freyja/tests/test_deconv.py

@@ -10,7 +10,7 @@
 
 class DeconvTests(unittest.TestCase):
     def test_buildLineageMap(self):
-        mapDict = buildLineageMap()
+        mapDict = buildLineageMap('-1')
         self.assertTrue('Alpha' == mapDict['B.1.1.7'])
         self.assertTrue('Delta' == mapDict['AY.4'])
 
@@ -32,7 +32,7 @@ def test_build_mix_and_depth_plus_reindex(self):
         self.assertFalse('20C' in df_barcodes.columns)
 
     def test_constellation_mapping(self):
-        mapDict = buildLineageMap()
+        mapDict = buildLineageMap('-1')
         vals = [0.1, 0.5, 0.31, 0.01, 0.02, 0.01]
         strains = ['B.1.617.2', 'Q.3', 'B.1.427', 'A.2.5', 'B.1.1', 'B.1.1.7']
         locDict = map_to_constellation(strains, vals, mapDict)
@@ -61,7 +61,7 @@ def test_demixing(self):
             abundances[sample_strains.tolist().index(strain2)], mixFracs[1])
 
     def test_boot(self):
-        mapDict = buildLineageMap()
+        mapDict = buildLineageMap('-1')
         df_barcodes = pd.read_csv('freyja/data/usher_barcodes.csv',
                                   index_col=0)
         muts = list(df_barcodes.columns)