switch to prot ksize=57

sourmash-bio · Jun 12, 2020 · 20d5cf5 · 20d5cf5
1 parent ed36010
commit 20d5cf5
Show file tree

Hide file tree

Showing 14 changed files with 18 additions and 15 deletions.
diff --git a/tests/test-data/prot/build.sh b/tests/test-data/prot/build.sh
@@ -0,0 +1,3 @@
+sourmash compute -k 57 *_protein.faa.gz --scaled=100 -f --input-is-protein  --no-dna --outdir protein --protein
+sourmash compute -k 57 *_protein.faa.gz --scaled=100 -f --input-is-protein  --no-dna --outdir hp --hp
+sourmash compute -k 57 *_protein.faa.gz --scaled=100 -f --input-is-protein  --no-dna --outdir dayhoff --dayhoff
diff --git a/tests/test-data/prot/dayhoff.lca.json.gz b/tests/test-data/prot/dayhoff.lca.json.gz
diff --git a/tests/test-data/prot/dayhoff.sbt.zip b/tests/test-data/prot/dayhoff.sbt.zip
diff --git a/tests/test-data/prot/dayhoff/GCA_001593925.1_ASM159392v1_protein.faa.gz.sig b/tests/test-data/prot/dayhoff/GCA_001593925.1_ASM159392v1_protein.faa.gz.sig
diff --git a/tests/test-data/prot/dayhoff/GCA_001593935.1_ASM159393v1_protein.faa.gz.sig b/tests/test-data/prot/dayhoff/GCA_001593935.1_ASM159393v1_protein.faa.gz.sig
diff --git a/tests/test-data/prot/hp.lca.json.gz b/tests/test-data/prot/hp.lca.json.gz
diff --git a/tests/test-data/prot/hp.sbt.zip b/tests/test-data/prot/hp.sbt.zip
diff --git a/tests/test-data/prot/hp/GCA_001593925.1_ASM159392v1_protein.faa.gz.sig b/tests/test-data/prot/hp/GCA_001593925.1_ASM159392v1_protein.faa.gz.sig
diff --git a/tests/test-data/prot/hp/GCA_001593935.1_ASM159393v1_protein.faa.gz.sig b/tests/test-data/prot/hp/GCA_001593935.1_ASM159393v1_protein.faa.gz.sig
diff --git a/tests/test-data/prot/protein.lca.json.gz b/tests/test-data/prot/protein.lca.json.gz
diff --git a/tests/test-data/prot/protein.sbt.zip b/tests/test-data/prot/protein.sbt.zip
diff --git a/tests/test-data/prot/protein/GCA_001593925.1_ASM159392v1_protein.faa.gz.sig b/tests/test-data/prot/protein/GCA_001593925.1_ASM159392v1_protein.faa.gz.sig
diff --git a/tests/test-data/prot/protein/GCA_001593935.1_ASM159393v1_protein.faa.gz.sig b/tests/test-data/prot/protein/GCA_001593935.1_ASM159393v1_protein.faa.gz.sig
diff --git a/tests/test_lca.py b/tests/test_lca.py
@@ -1858,7 +1858,7 @@ def test_lca_db_protein_build():
     sig1 = sourmash.load_one_signature(sigfile1)
     sig2 = sourmash.load_one_signature(sigfile2)
 
-    db = sourmash.lca.LCA_Database(ksize=19, scaled=100, moltype='protein')
+    db = sourmash.lca.LCA_Database(ksize=57, scaled=100, moltype='protein')
     assert db.insert(sig1)
     assert db.insert(sig2)
 
@@ -1885,7 +1885,7 @@ def test_lca_db_protein_save_load(c):
     sig1 = sourmash.load_one_signature(sigfile1)
     sig2 = sourmash.load_one_signature(sigfile2)
 
-    db = sourmash.lca.LCA_Database(ksize=19, scaled=100, moltype='protein')
+    db = sourmash.lca.LCA_Database(ksize=57, scaled=100, moltype='protein')
     assert db.insert(sig1)
     assert db.insert(sig2)
 
@@ -1921,7 +1921,7 @@ def test_lca_db_protein_command_index(c):
 
     c.run_sourmash('lca', 'index', lineages, db_out, sigfile1, sigfile2,
                    '-C', '3', '--split-identifiers', '--require-taxonomy',
-                   '--scaled', '100', '-k', '19', '--protein')
+                   '--scaled', '100', '-k', '57', '--protein')
 
     x = sourmash.lca.lca_db.load_single_database(db_out)
     db2 = x[0]
@@ -1967,7 +1967,7 @@ def test_lca_db_hp_build():
     sig1 = sourmash.load_one_signature(sigfile1)
     sig2 = sourmash.load_one_signature(sigfile2)
 
-    db = sourmash.lca.LCA_Database(ksize=19, scaled=1, moltype='hp')
+    db = sourmash.lca.LCA_Database(ksize=57, scaled=100, moltype='hp')
     assert db.insert(sig1)
     assert db.insert(sig2)
 
@@ -1994,7 +1994,7 @@ def test_lca_db_hp_save_load(c):
     sig1 = sourmash.load_one_signature(sigfile1)
     sig2 = sourmash.load_one_signature(sigfile2)
 
-    db = sourmash.lca.LCA_Database(ksize=19, scaled=1, moltype='hp')
+    db = sourmash.lca.LCA_Database(ksize=57, scaled=100, moltype='hp')
     assert db.insert(sig1)
     assert db.insert(sig2)
 
@@ -2030,7 +2030,7 @@ def test_lca_db_hp_command_index(c):
 
     c.run_sourmash('lca', 'index', lineages, db_out, sigfile1, sigfile2,
                    '-C', '3', '--split-identifiers', '--require-taxonomy',
-                   '--scaled', '1', '-k', '19', '--hp')
+                   '--scaled', '100', '-k', '57', '--hp')
 
     x = sourmash.lca.lca_db.load_single_database(db_out)
     db2 = x[0]
@@ -2076,7 +2076,7 @@ def test_lca_db_dayhoff_build():
     sig1 = sourmash.load_one_signature(sigfile1)
     sig2 = sourmash.load_one_signature(sigfile2)
 
-    db = sourmash.lca.LCA_Database(ksize=19, scaled=100, moltype='dayhoff')
+    db = sourmash.lca.LCA_Database(ksize=57, scaled=100, moltype='dayhoff')
     assert db.insert(sig1)
     assert db.insert(sig2)
 
@@ -2103,7 +2103,7 @@ def test_lca_db_dayhoff_save_load(c):
     sig1 = sourmash.load_one_signature(sigfile1)
     sig2 = sourmash.load_one_signature(sigfile2)
 
-    db = sourmash.lca.LCA_Database(ksize=19, scaled=100, moltype='dayhoff')
+    db = sourmash.lca.LCA_Database(ksize=57, scaled=100, moltype='dayhoff')
     assert db.insert(sig1)
     assert db.insert(sig2)
 
@@ -2139,7 +2139,7 @@ def test_lca_db_dayhoff_command_index(c):
 
     c.run_sourmash('lca', 'index', lineages, db_out, sigfile1, sigfile2,
                    '-C', '3', '--split-identifiers', '--require-taxonomy',
-                   '--scaled', '100', '-k', '19', '--dayhoff')
+                   '--scaled', '100', '-k', '57', '--dayhoff')
 
     x = sourmash.lca.lca_db.load_single_database(db_out)
     db2 = x[0]