|
a |
|
b/qiita_db/reference.py |
|
|
1 |
# ----------------------------------------------------------------------------- |
|
|
2 |
# Copyright (c) 2014--, The Qiita Development Team. |
|
|
3 |
# |
|
|
4 |
# Distributed under the terms of the BSD 3-clause License. |
|
|
5 |
# |
|
|
6 |
# The full license is in the file LICENSE, distributed with this software. |
|
|
7 |
# ----------------------------------------------------------------------------- |
|
|
8 |
from os.path import join |
|
|
9 |
|
|
|
10 |
import qiita_db as qdb |
|
|
11 |
|
|
|
12 |
|
|
|
13 |
class Reference(qdb.base.QiitaObject): |
|
|
14 |
r"""Object to interact with reference sequence databases |
|
|
15 |
|
|
|
16 |
Attributes |
|
|
17 |
---------- |
|
|
18 |
sequence_fp |
|
|
19 |
taxonomy_fp |
|
|
20 |
tree_fp |
|
|
21 |
|
|
|
22 |
Methods |
|
|
23 |
------- |
|
|
24 |
create |
|
|
25 |
exists |
|
|
26 |
|
|
|
27 |
See Also |
|
|
28 |
-------- |
|
|
29 |
QiitaObject |
|
|
30 |
""" |
|
|
31 |
_table = "reference" |
|
|
32 |
|
|
|
33 |
@classmethod |
|
|
34 |
def create(cls, name, version, seqs_fp, tax_fp=None, tree_fp=None): |
|
|
35 |
r"""Creates a new reference object with a new id on the storage system |
|
|
36 |
|
|
|
37 |
Parameters |
|
|
38 |
---------- |
|
|
39 |
name : str |
|
|
40 |
The name of the reference database |
|
|
41 |
version : str |
|
|
42 |
The version of the reference database |
|
|
43 |
seqs_fp : str |
|
|
44 |
The path to the reference sequence file |
|
|
45 |
tax_fp : str, optional |
|
|
46 |
The path to the reference taxonomy file |
|
|
47 |
tree_fp : str, optional |
|
|
48 |
The path to the reference tree file |
|
|
49 |
|
|
|
50 |
Returns |
|
|
51 |
------- |
|
|
52 |
A new instance of `cls` to access to the Reference stored in the DB |
|
|
53 |
|
|
|
54 |
Raises |
|
|
55 |
------ |
|
|
56 |
QiitaDBDuplicateError |
|
|
57 |
If the reference database with name `name` and version `version` |
|
|
58 |
already exists on the system |
|
|
59 |
""" |
|
|
60 |
with qdb.sql_connection.TRN: |
|
|
61 |
if cls.exists(name, version): |
|
|
62 |
raise qdb.exceptions.QiitaDBDuplicateError( |
|
|
63 |
"Reference", "Name: %s, Version: %s" % (name, version)) |
|
|
64 |
|
|
|
65 |
fps = [(seqs_fp, |
|
|
66 |
qdb.util.convert_to_id("reference_seqs", "filepath_type"))] |
|
|
67 |
seq_id = qdb.util.insert_filepaths( |
|
|
68 |
fps, "%s_%s" % (name, version), "reference")[0] |
|
|
69 |
|
|
|
70 |
# Check if the database has taxonomy file |
|
|
71 |
tax_id = None |
|
|
72 |
if tax_fp: |
|
|
73 |
fps = [ |
|
|
74 |
(tax_fp, |
|
|
75 |
qdb.util.convert_to_id("reference_tax", "filepath_type"))] |
|
|
76 |
tax_id = qdb.util.insert_filepaths( |
|
|
77 |
fps, "%s_%s" % (name, version), "reference")[0] |
|
|
78 |
|
|
|
79 |
# Check if the database has tree file |
|
|
80 |
tree_id = None |
|
|
81 |
if tree_fp: |
|
|
82 |
fps = [ |
|
|
83 |
(tree_fp, |
|
|
84 |
qdb.util.convert_to_id("reference_tree", "filepath_type")) |
|
|
85 |
] |
|
|
86 |
tree_id = qdb.util.insert_filepaths( |
|
|
87 |
fps, "%s_%s" % (name, version), "reference")[0] |
|
|
88 |
|
|
|
89 |
# Insert the actual object to the db |
|
|
90 |
sql = """INSERT INTO qiita.{0} |
|
|
91 |
(reference_name, reference_version, sequence_filepath, |
|
|
92 |
taxonomy_filepath, tree_filepath) |
|
|
93 |
VALUES (%s, %s, %s, %s, %s) |
|
|
94 |
RETURNING reference_id""".format(cls._table) |
|
|
95 |
qdb.sql_connection.TRN.add( |
|
|
96 |
sql, [name, version, seq_id, tax_id, tree_id]) |
|
|
97 |
id_ = qdb.sql_connection.TRN.execute_fetchlast() |
|
|
98 |
|
|
|
99 |
return cls(id_) |
|
|
100 |
|
|
|
101 |
@classmethod |
|
|
102 |
def exists(cls, name, version): |
|
|
103 |
r"""Checks if a given object info is already present on the DB |
|
|
104 |
|
|
|
105 |
Parameters |
|
|
106 |
---------- |
|
|
107 |
name : str |
|
|
108 |
The name of the reference database |
|
|
109 |
version : str |
|
|
110 |
The version of the reference database |
|
|
111 |
|
|
|
112 |
Raises |
|
|
113 |
------ |
|
|
114 |
QiitaDBNotImplementedError |
|
|
115 |
If the method is not overwritten by a subclass |
|
|
116 |
""" |
|
|
117 |
with qdb.sql_connection.TRN: |
|
|
118 |
sql = """SELECT EXISTS( |
|
|
119 |
SELECT * FROM qiita.{0} |
|
|
120 |
WHERE reference_name=%s |
|
|
121 |
AND reference_version=%s)""".format(cls._table) |
|
|
122 |
qdb.sql_connection.TRN.add(sql, [name, version]) |
|
|
123 |
return qdb.sql_connection.TRN.execute_fetchlast() |
|
|
124 |
|
|
|
125 |
@property |
|
|
126 |
def name(self): |
|
|
127 |
with qdb.sql_connection.TRN: |
|
|
128 |
sql = """SELECT reference_name FROM qiita.{0} |
|
|
129 |
WHERE reference_id = %s""".format(self._table) |
|
|
130 |
qdb.sql_connection.TRN.add(sql, [self._id]) |
|
|
131 |
return qdb.sql_connection.TRN.execute_fetchlast() |
|
|
132 |
|
|
|
133 |
@property |
|
|
134 |
def version(self): |
|
|
135 |
with qdb.sql_connection.TRN: |
|
|
136 |
sql = """SELECT reference_version FROM qiita.{0} |
|
|
137 |
WHERE reference_id = %s""".format(self._table) |
|
|
138 |
qdb.sql_connection.TRN.add(sql, [self._id]) |
|
|
139 |
return qdb.sql_connection.TRN.execute_fetchlast() |
|
|
140 |
|
|
|
141 |
def _retrieve_filepath(self, column): |
|
|
142 |
def path_builder(db_dir, filepath, mountpoint, subdirectory, obj_id): |
|
|
143 |
if subdirectory: |
|
|
144 |
return join(db_dir, mountpoint, str(obj_id), filepath) |
|
|
145 |
else: |
|
|
146 |
return join(db_dir, mountpoint, filepath) |
|
|
147 |
|
|
|
148 |
with qdb.sql_connection.TRN: |
|
|
149 |
sql = """SELECT filepath, mountpoint, subdirectory |
|
|
150 |
FROM qiita.filepath f |
|
|
151 |
JOIN qiita.filepath_type USING (filepath_type_id) |
|
|
152 |
JOIN qiita.data_directory USING (data_directory_id) |
|
|
153 |
JOIN qiita.{0} r ON r.{1} = f.filepath_id |
|
|
154 |
WHERE r.reference_id = %s""".format(self._table, column) |
|
|
155 |
qdb.sql_connection.TRN.add(sql, [self._id]) |
|
|
156 |
result = qdb.sql_connection.TRN.execute_fetchindex() |
|
|
157 |
if result: |
|
|
158 |
# If results is not empty, there will be only 1 result in the |
|
|
159 |
# list, hence using the 0 index |
|
|
160 |
fp, mountpoint, subdir = result[0] |
|
|
161 |
db_dir = qdb.util.get_db_files_base_dir() |
|
|
162 |
return path_builder(db_dir, fp, mountpoint, subdir, self._id) |
|
|
163 |
else: |
|
|
164 |
return '' |
|
|
165 |
|
|
|
166 |
@property |
|
|
167 |
def sequence_fp(self): |
|
|
168 |
return self._retrieve_filepath('sequence_filepath') |
|
|
169 |
|
|
|
170 |
@property |
|
|
171 |
def taxonomy_fp(self): |
|
|
172 |
return self._retrieve_filepath('taxonomy_filepath') |
|
|
173 |
|
|
|
174 |
@property |
|
|
175 |
def tree_fp(self): |
|
|
176 |
return self._retrieve_filepath('tree_filepath') |