ENH Set macrel/{__version__} as user agent

BigDataBiology · Jun 28, 2024 · 3fd5e41 · 3fd5e41
1 parent 0ab399b
commit 3fd5e41
Show file tree

Hide file tree

Showing 12 changed files with 21 additions and 16 deletions.
diff --git a/macrel/ampsphere.py b/macrel/ampsphere.py
@@ -5,6 +5,11 @@
 import logging
 
 from macrel.utils import open_output
+from macrel.macrel_version import __version__
+
+REQUESTS_HEADER = {
+        'User-Agent': f'macrel/{__version__} (python-requests)'
+        }
 
 def get_cache_directory(args):
     '''Get cache directory'''
@@ -40,7 +45,7 @@ def maybe_download_ampsphere_mmseqs(args):
     AMPSPHERE_MMSEQS2_URL = 'https://ampsphere-api.big-data-biology.org/v1/downloads/AMPSphere_latest.mmseqsdb.tar.xz'
     with tempfile.TemporaryDirectory() as tmpdir:
         tfile = path.join(tmpdir, 'AMPSphere_latest.mmseqsdb.tar.xz')
-        r = requests.get(AMPSPHERE_MMSEQS2_URL, stream=True)
+        r = requests.get(AMPSPHERE_MMSEQS2_URL, stream=True, headers=REQUESTS_HEADER)
         with open(tfile, 'wb') as f:
             for chunk in r.iter_content(chunk_size=8192):
                 f.write(chunk)
@@ -111,7 +116,7 @@ def maybe_download_ampsphere_faa(args):
     if args.no_download_database:
         return None
     URL = 'https://zenodo.org/records/6511404/files/AMPSphere_v.2022-03.faa.gz?download=1'
-    r = requests.get(URL, stream=True)
+    r = requests.get(URL, stream=True, headers=REQUESTS_HEADER)
     with open_output(target, 'wb') as f:
         for chunk in r.iter_content(chunk_size=8192):
             f.write(chunk)
@@ -144,15 +149,15 @@ def get_ampsphere_exact_match_local(args, seqs):
 def get_ampsphere_exact_match(seq, query_name):
     '''Get exact match from AMPSphere API'''
     URL = f'https://ampsphere-api.big-data-biology.org/v1/search/sequence-match?query={seq}'
-    response = requests.get(URL)
+    response = requests.get(URL, headers=REQUESTS_HEADER)
     data = response.json()
     return pd.DataFrame.from_dict({query_name : data}, orient='index')
 
 def get_ampsphere_mmseqs_match(seq, query_name):
     '''Get MMSeqs2 match from AMPSphere API'''
     query = f'>{query_name}\n{seq}'
     URL = f'https://ampsphere-api.big-data-biology.org/v1/search/mmseqs?query={query}'
-    response = requests.get(URL)
+    response = requests.get(URL, headers=REQUESTS_HEADER)
     data = response.json()
     return pd.DataFrame.from_dict(data)\
             .drop("alignment_strings", axis=1)\
@@ -162,7 +167,7 @@ def get_ampsphere_hmmer_match(seq, query_name):
     '''Get HMMER match from AMPSphere API'''
     query = f'>{query_name}\n{seq}'
     URL = f'https://ampsphere-api.big-data-biology.org/v1/search/hmmer?query={query}'
-    response = requests.get(URL)
+    response = requests.get(URL, headers=REQUESTS_HEADER)
     data = response.json()
     if not data:
         return pd.DataFrame()

diff --git a/macrel/macrel_version.py b/macrel/macrel_version.py
@@ -1 +1 @@
-__version__ = '1.4.0'
+__version__ = '1.5.0.dev0'
diff --git a/tests/contigs.cluster/expected.prediction b/tests/contigs.cluster/expected.prediction
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 Access	Sequence	AMP_family	AMP_probability	Hemolytic	Hemolytic_probability
 smORF_2	RFLIKMVKVNLMNGKLIRKISLM	CLP	0.634	Hemo	0.871
 smORF_19	FFNDGKGTIYYGIKKYFRIYF	CLP	0.673	Hemo	0.822
diff --git a/tests/contigs.nosmorfs/expected.percontigs b/tests/contigs.nosmorfs/expected.percontigs
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 # Macrel calculated for the sample a density of 0.000 AMPs / Mbp.
 contig	length	ORFs	smORFs	AMPs
 scaffold2530_2_MH0058	1324	1	0	0
diff --git a/tests/contigs.nosmorfs/expected.prediction b/tests/contigs.nosmorfs/expected.prediction
@@ -1,2 +1,2 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 Access	Sequence	AMP_family	AMP_probability	Hemolytic	Hemolytic_probability
diff --git a/tests/contigs/expected.percontigs b/tests/contigs/expected.percontigs
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 # Macrel calculated for the sample a density of 45.062 AMPs / Mbp.
 contig	length	ORFs	smORFs	AMPs
 C4060843_1_MH0058	518	1	1	0

diff --git a/tests/contigs/expected.prediction b/tests/contigs/expected.prediction
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 Access	Sequence	AMP_family	AMP_probability	Hemolytic	Hemolytic_probability
 scaffold75334_1_MH0058_1	RFLIKMVKVNLMNGKLIRKISLM	CLP	0.634	Hemo	0.871
 scaffold33693_17_MH0058_2	FFNDGKGTIYYGIKKYFRIYF	CLP	0.673	Hemo	0.822
diff --git a/tests/peptides/expected.prediction b/tests/peptides/expected.prediction
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 Access	Sequence	AMP_family	AMP_probability	Hemolytic	Hemolytic_probability
 AP00002|AMP	YVPLPNVPQPGRRPFPTFPGQGPFNPKIKWPQGY	CLP	0.861	Hemo	0.663
 AP00007|AMP	GNNRPVYIPQPRPPHPRL	CLP	0.970	Hemo	0.515

diff --git a/tests/reads.se/expected.percontigs b/tests/reads.se/expected.percontigs
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 # Macrel calculated for the sample a density of 59.743 AMPs / Mbp.
 contig	length	ORFs	smORFs	AMPs
 k47_0	3379	4	2	0

diff --git a/tests/reads.se/expected.prediction b/tests/reads.se/expected.prediction
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 Access	Sequence	AMP_family	AMP_probability	Hemolytic	Hemolytic_probability
 k47_10_1	RFLIKMVKVNLMNGKLIRKISLM	CLP	0.634	Hemo	0.871
 k47_11_1	FFNDGKGTIYYGIKKYFRIYF	CLP	0.673	Hemo	0.822
diff --git a/tests/reads/expected.percontigs b/tests/reads/expected.percontigs
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 # Macrel calculated for the sample a density of 57.627 AMPs / Mbp.
 contig	length	ORFs	smORFs	AMPs
 k77_11	1303	2	1	0

diff --git a/tests/reads/expected.prediction b/tests/reads/expected.prediction
@@ -1,4 +1,4 @@
-# Prediction from macrel v1.4.0
+# Prediction from macrel v1.5.0.dev0
 Access	Sequence	AMP_family	AMP_probability	Hemolytic	Hemolytic_probability
 k77_12_1	RFLIKMVKVNLMNGKLIRKISLM	CLP	0.634	Hemo	0.871
 k77_15_1	FFNDGKGTIYYGIKKYFRIYF	CLP	0.673	Hemo	0.822