db.py 14.3 KB
Newer Older
aknecht2's avatar
aknecht2 committed
1
from pymongo import MongoClient
2
import pymongo.errors
3
4
5
import gridfs
import sys
import traceback
6
import os
7
import itertools
8
import chipathlon.conf
9
from pprint import pprint
aknecht2's avatar
aknecht2 committed
10

11

12
13
14
class MongoDB(object):

    def __init__(self, host, username, password):
15
16
        self.client = MongoClient(host)
        self.db = self.client.chipseq
17
        try:
18
            self.db.authenticate(username, password, mechanism="SCRAM-SHA-1")
19
20
21
22
23
24
25
        except:
            print("Could not authenticate to db %s!" % (host,))
            print traceback.format_exc()
            sys.exit(1)
        self.gfs = gridfs.GridFS(self.db)
        return

26
27
28
29
30
31
32
33
34
35
36
37
    def delete_result(self, result_id):
        # Make sure result exists
        cursor = self.db.results.find({
            "_id": result_id
        })
        if cursor.count() == 1:
            result = cursor.next()
            self.gfs.delete(result["gridfs_id"])
            self.db[result["result_type"]].delete_many({"result_id": result["_id"]})
            self.db.results.delete_one({"_id": result["_id"]})
        else:
            print "result_id %s doesn't exist." % (result_id,)
38
39
        return

40
    def create_result(self, output_file, control_sample_ids, experiment_sample_ids, result_type, additional_data = {}, gfs_attributes = {}):
41
42
        # Make sure output_file exists
        if os.path.isfile(output_file):
43
            # Make sure that all control_sample_ids & experiment_sample_ids are valid
44
            # REMEMBER, these are ids for control & experiment SAMPLES
45
46
            valid_controls = [self.is_valid_sample(cid) for cid in control_sample_ids]
            valid_experiments = [self.is_valid_sample(eid) for eid in experiment_sample_ids]
47
48
49
50
51
52
53
54
            if all(valid_controls) and all(valid_experiments):
                # First, we load the output file into gfs
                with open(output_file, "r") as rh:
                    # Calling put returns the gfs id
                    gridfs_id = self.gfs.put(rh, filename=os.path.basename(output_file), **gfs_attributes)
                # Now, we create the actual result entry by combining all necessary info
                result_entry = {
                    "gridfs_id": gridfs_id,
55
56
                    "control_sample_ids": control_sample_ids,
                    "experiment_sample_ids": experiment_sample_ids,
57
58
59
60
61
62
63
64
                    "result_type": result_type
                }
                # Add additional attributes into the result_entry
                result_entry.update(additional_data)
                # Insert the entry into the database, and return the id
                result = self.db.results.insert_one(result_entry)
                return (True, "Result created successfully.", result.inserted_id)
            else:
65
                msg = "Not all input ids are valid.  The following are invalid: "
66
                for id_list, valid_list in zip([control_sample_ids, experiment_sample_ids], [valid_controls, valid_experiments]):
67
                    msg += ", ".join([id_list[i] for i, valid in enumerate(valid_list) if not valid])
68
69
70
71
        else:
            msg = "Specified output_file %s does not exist." % (output_file,)
        return (False, msg, None)

72
73
74
75
76
77
78
    def save_bam(self, bam_file, control_sample_ids, experiment_sample_ids, additional_data = {}):
        # Create result entry for bam files.  Since bam is a binary format, the file will only
        # be stored in GridFS
        valid, msg, result_id = self.create_result(bam_file, control_sample_ids, experiment_sample_ids, "bam", additional_data, gfs_attributes = {"file_type": "bam"})
        return (valid, msg, result_id)


79
    def save_bed(self, bed_file, control_sample_ids, experiment_sample_ids, additional_data = {}):
80
        # Create result_entry for bed_file
81
        valid, msg, result_id = self.create_result(bed_file, control_sample_ids, experiment_sample_ids, "bed", additional_data, gfs_attributes = {"file_type": "bed"})
82
83
84
85
86
        if valid:
            # Now we load the actual bed data into the bed collection.
            # Data is in a six column format
            # chr, start, end, name, score, strand
            # Load data using a list comprehension over lines,
87
88
89
            # then insert with insert_one()
            # Each document contains "n_lines" number of lines from the
            # result BED file.
90
            print "loading bed_data..."
91
            with open(bed_file, "r") as rh:
92
                msg = "Bed file successfully inserted."
93
94
                # Lazy load files in specified line chunk size
                n_lines = chipathlon.conf.result_lines_per_document
95
96
97
                line_set = list(itertools.islice(rh, n_lines))
                while line_set:
                    try:
98
99
100
101
                        result_lines = []
                        for line in line_set:
                            line_info = line.split()
                            line_record =  {
102
103
104
105
106
                                "chr": line_info[0],
                                "start": line_info[1],
                                "end": line_info[2],
                                "name": line_info[3],
                                "score": line_info[4],
107
                                "strand": line_info[5],
108
                            }
109
110
111
112
                            result_lines.append(line_record)

                        self.db.bed.insert_one({"result_id": result_id, "result_lines": result_lines})

113
114
115
116
117
                    except pymongo.errors.OperationFailure as e:
                        valid = False
                        msg = "Error inserting bed_file %s: %s" % (bed_file, e)
                    line_set = list(itertools.islice(rh, n_lines))
        return (valid, msg, result_id)
118

119
    def save_peak(self, peak_file, control_sample_ids, experiment_sample_ids, additional_data = {}):
120
        # Create result_entry for peak_file
121
        valid, msg, result_id = self.create_result(peak_file, control_sample_ids, experiment_sample_ids, "peak", additional_data, gfs_attributes = {"file_type": os.path.splitext(peak_file)[1][1:]})
122
123
124
125
        if valid:
            # Now we load the actual peak data into the collection
            # Data is in a 10 column format
            # chr, start, end, name, score, strand, signal_value, p_value, q_value, summit
126
127
            # Each document contains "n_lines" number of lines from the
            # result peak file.
128
            with open(peak_file, "r") as rh:
129
                msg = "Peak file successfully inserted."
130
131
                # Lazy load files in specified line chunk size
                n_lines = chipathlon.conf.result_lines_per_document
132
133
134
                line_set = list(itertools.islice(rh, n_lines))
                while line_set:
                    try:
135
136
137
138
                        result_lines = []
                        for line in line_set:
                            line_info = line.split()
                            line_record =  {
139
140
141
142
143
144
145
146
147
148
149
                                "chr": line_info[0],
                                "start": line_info[1],
                                "end": line_info[2],
                                "name": line_info[3],
                                "score": line_info[4],
                                "strand": line_info[5],
                                "signal_value": line_info[6],
                                "p_value": line_info[7],
                                "q_value": line_info[8],
                                "summit": line_info[9]
                            }
150
151
152
153
                            result_lines.append(line_record)

                        self.db.peak.insert_one({"result_id": result_id, "result_lines": result_lines})

154
155
156
157
158
                    except pymongo.errors.OperationFailure as e:
                        valid = False
                        msg = "Error inserting peak_file %s: %s" % (peak_file, e)
                    line_set = list(itertools.islice(rh, n_lines))
        return (valid, msg, result_id)
159

160
161
162
163
164
165
166
167
168
169
170
    def is_valid_sample(self, sample_accession):
        try:
            cursor = self.db.samples.find({
                "accession": sample_accession
            })
            if cursor.count() == 1:
                return True
        except pymongo.errors.OperationFailure as e:
            print "Error with sample_accession %s: %s" % (sample_accession, e)
        return False

171
172
173
174
175
176
177
178
179
180
181
182
183
    def is_valid_experiment(self, experiment_id):
        try:
            cursor = self.db.experiments.find({
                "target": {"$exists": True},
                "revoked_files.0": {"$exists": False},
                "@id": "/experiments/%s/" % (experiment_id,)
            })
            if cursor.count() == 1:
                return True
        except pymongo.errors.OperationFailure as e:
            print "Error with experiment_id %s: %s" % (experiment_id, e)
        return False

184
185
186
187
188
189
190
191
192
193
194
195
    def check_valid_samples(self):
        cursor = self.db.experiments.aggregate([
            {
                "$match": {
                    "target": {"$exists": True},
                    "revoked_files.0": {"$exists": False},
                    "assembly.0": {"$exists": True},
                    "assembly.1": {"$exists": False}
                }
            },
            {
                "$lookup": {
196
                    "from": "samples",
197
                    "localField": "uuid",
198
199
200
201
202
203
204
205
206
207
208
209
210
                    "foreignField": "experiment_id",
                    "as": "samples"
                }
            }
        ])
        total = 0
        has_samples = 0
        for document in cursor:
            total += 1
            if len(document["samples"]) > 0:
                has_samples += 1
        return (has_samples, total)

211
212
213
214
215
216
217
218
219
220
221
222
223
224
    def get_assembly(self, experiment_id):
        valid = True
        msg = ""
        data = ""
        cursor = self.db.experiments.find({
            "target": {"$exists": True},
            "revoked_files.0": {"$exists": False},
            "assembly.0": {"$exists": True},
            "assembly.1": {"$exists": False},
            "@id": "/experiments/%s/" % (experiment_id,)
        })
        if cursor.count() == 1:
            document = cursor.next()
            data = document["assembly"][0]
aknecht2's avatar
aknecht2 committed
225
            msg = "Succesfully retrieved assembly for experiment with id '%s'.\n" % (experiment_id,)
226
227
        else:
            valid = False
aknecht2's avatar
aknecht2 committed
228
            msg = "Experiment with id '%s' does not exist.\n" % (experiment_id,)
229
230
        return (valid, msg, data)

231
232
233
234
    def get_samples(self, experiment_id):
        valid = True
        msg = ""
        data = {}
235
236
237
238
        # First, check to make sure the target experiment is valid
        if self.is_valid_experiment(experiment_id):
            # Next, we check that there is a least 1 possible control
            check3 = self.db.experiments.find({
239
240
                "target": {"$exists": True},
                "revoked_files.0": {"$exists": False},
241
242
243
                "assembly.0": {"$exists": True},
                "assembly.1": {"$exists": False},
                "possible_controls.0": {"$exists": True},
244
245
                "@id": "/experiments/%s/" % (experiment_id,)
            })
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
            if check3.count() == 1:
                # Complicated aggregtaion pipeline does the following steps:
                # 1. Find the experiment that matches the given id
                # 2. Join samples into the collection by exp_id
                # 3. Iterate through possible_controls
                # 4. Join possible_control data into control_exps
                # 5. Iterate through control_exps
                # 6. Join samples into the control_exps by exp_id
                # 7. Re-aggregate all data into arrays
                pipeline = [
                    {
                        "$match": {
                            "target": {"$exists": True},
                            "revoked_files.0": {"$exists": False},
                            "assembly.0": {"$exists": True},
                            "assembly.1": {"$exists": False},
                            "possible_controls.0": {"$exists": True},
                            "@id": "/experiments/%s/" % (experiment_id,)
264
                        }
265
266
267
268
269
270
271
                    },
                    {
                        "$lookup": {
                            "from": "samples",
                            "localField": "uuid",
                            "foreignField": "experiment_id",
                            "as": "samples"
272
                        }
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
                    },
                    {
                        "$unwind": "$possible_controls"
                    },
                    {
                        "$lookup": {
                            "from": "samples",
                            "localField": "possible_controls.uuid",
                            "foreignField": "experiment_id",
                            "as": "possible_controls.samples"
                        }
                    },
                    {
                        "$group": {
                            "_id": "$_id",
                            "possible_controls": {"$push": "$possible_controls"},
                            "samples": {"$push": "$samples"}
                        }
                    }
                ]
                cursor = self.db.experiments.aggregate(pipeline)
                # We should have only 1 document
                document = cursor.next()
                control_inputs = [sample for control in document["possible_controls"] for sample in control["samples"] if ("file_type" in sample and sample["file_type"] == "fastq")]
                experiment_inputs = [sample for sample in document["samples"][0] if ("file_type" in sample and sample["file_type"] == "fastq")]
                if (len(control_inputs) > 0 and len(experiment_inputs) > 0):
                    msg = "Succesfully retrieved input files for experiment with id '%s'.\n" % (experiment_id,)
                    data = {
                        "control": control_inputs,
                        "experiment": experiment_inputs
                    }
304
305
                else:
                    valid = False
306
                    msg = "Experiment with id '%s' has %s possible control inputs, and %s possible experiment inputs.\n" % (experiment_id, len(control_inputs), len(experiment_inputs))
307
308
            else:
                valid = False
309
                msg = "Experiment with id '%s' does not have possible_controls.\n" % (experiment_id,)
310
311
        else:
            valid = False
312
            msg = "Experiment with id '%s' is not valid!  It may not exist, or it may be missing required metadata.\n" % (experiment_id,)
313
        return (valid, msg, data)