models.py 35 KB
Newer Older
1
2
3
4
"""
Models used in iPPI-DB
"""

Hervé  MENAGER's avatar
Hervé MENAGER committed
5
from __future__ import unicode_literals
6
import operator
Hervé  MENAGER's avatar
Hervé MENAGER committed
7
8

from django.db import models
9
from django.conf import settings
10
from django.db.models import Max, Min, Count, F, Q, Case, When, Sum
11
from django.db.models.functions import Cast
12
from django.db.models import FloatField, IntegerField, BooleanField                                                                                                                                                                 
13

Hervé  MENAGER's avatar
Hervé MENAGER committed
14

15
from .utils import FingerPrinter, smi2inchi, smi2inchikey
16
from .ws import get_pubmed_info, get_google_patent_info, get_uniprot_info, get_taxonomy_info, get_go_info, get_pfam_info
17

Hervé  MENAGER's avatar
Hervé MENAGER committed
18

19
class AutoFillableModel(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
20

21
22
23
24
25
26
27
28
29
30
31
32
33
    """
    AutoFillableModel makes it possible to automatically fill model fields from
    external sources in the autofill() method
    The save method allows to either include autofill or not. in autofill kwarg is
    set to True, save() will first call autofill(), otherwise it won't
    """

    class Meta:
        abstract = True

    def save(self, *args, **kwargs):
        if kwargs.get('autofill') is True:
            self.autofill()
34
35
        if 'autofill' in kwargs:
            del kwargs['autofill']
Hervé  MENAGER's avatar
Hervé MENAGER committed
36
        super(AutoFillableModel, self).save(*args, **kwargs)
37
38
39


class Bibliography(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
40
    """
Hervé  MENAGER's avatar
Hervé MENAGER committed
41
42
    Bibliography references
    (publications or patents)
Hervé  MENAGER's avatar
Hervé MENAGER committed
43
44
    """
    SOURCES = (
45
46
47
        ('PM', 'PubMed ID'),
        ('PT', 'Patent'),
        ('DO', 'DOI ID')
Hervé  MENAGER's avatar
Hervé MENAGER committed
48
    )
Hervé  MENAGER's avatar
Hervé MENAGER committed
49
50
    source = models.CharField(
        'Bibliographic type', max_length=2, choices=SOURCES, default='PM')
Hervé  MENAGER's avatar
Hervé MENAGER committed
51
52
    id_source = models.CharField('Bibliographic ID', max_length=25)
    title = models.CharField('Title', max_length=300)
53
    journal_name = models.CharField('Journal name', max_length=50, null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
54
55
    authors_list = models.CharField('Authors list', max_length=500)
    biblio_year = models.PositiveSmallIntegerField('Year')
56
    cytotox = models.BooleanField('Cytotoxicity data', default=False)
Rachel TORCHET's avatar
Rachel TORCHET committed
57
58
59
60
    in_silico = models.BooleanField('in silico study', default=False)
    in_vitro = models.BooleanField('in vitro study', default=False)
    in_vivo = models.BooleanField('in vivo study', default=False)
    in_cellulo = models.BooleanField('in cellulo study', default=False)
Hervé  MENAGER's avatar
Hervé MENAGER committed
61
62
    pharmacokinetic = models.BooleanField(
        'pharmacokinetic study', default=False)
Rachel TORCHET's avatar
Rachel TORCHET committed
63
    xray = models.BooleanField('X-Ray data', default=False)
Hervé  MENAGER's avatar
Hervé MENAGER committed
64

65
    def autofill(self):
Hervé  MENAGER's avatar
Hervé MENAGER committed
66
67
68
69
        """
        fetch information from external services
        (Pubmed or Google patents)
        """
70
71
72
        if self.source == 'PM':
            info = get_pubmed_info(self.id_source)
        else:
73
            info = get_google_patent_info(self.id_source)
74
75
76
77
        self.title = info['title']
        self.journal_name = info['journal_name']
        self.authors_list = info['authors_list']
        self.biblio_year = info['biblio_year']
Hervé  MENAGER's avatar
Hervé MENAGER committed
78

Hervé  MENAGER's avatar
Hervé MENAGER committed
79
80
81
    class Meta:
        verbose_name_plural = "bibliographies"

82
83
    def __str__(self):
        return '{}, {}'.format(self.source, self.id_source)
84

Hervé  MENAGER's avatar
Hervé MENAGER committed
85

86
class Taxonomy(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
87
88
89
90
    """
    Taxonomy IDs (from NCBI Taxonomy) 
    and the corresponding human-readable name
    """
Hervé  MENAGER's avatar
Hervé MENAGER committed
91
92
    taxonomy_id = models.DecimalField(
        'NCBI TaxID', unique=True, max_digits=9, decimal_places=0)
Hervé  MENAGER's avatar
Hervé MENAGER committed
93
    name = models.CharField('Organism name', max_length=200)
94

95
    def autofill(self):
Hervé  MENAGER's avatar
Hervé MENAGER committed
96
97
98
99
        """
        fetch information from external services
        (NCBI Entrez)
        """
100
101
102
        info = get_taxonomy_info(self.taxonomy_id)
        self.name = info['scientific_name']

103
104
105
    def __str__(self):
        return self.name

Hervé  MENAGER's avatar
Hervé MENAGER committed
106
107
    class Meta:
        verbose_name_plural = "taxonomies"
Hervé  MENAGER's avatar
Hervé MENAGER committed
108

Hervé  MENAGER's avatar
Hervé MENAGER committed
109

110
class MolecularFunction(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
111
112
113
114
    """
    Molecular functions (from Gene Ontology) 
    and the corresponding human-readable description
    """
Hervé  MENAGER's avatar
Hervé MENAGER committed
115
    go_id = models.CharField('Gene Ontology ID', unique=True, max_length=10)
Hervé  MENAGER's avatar
Hervé MENAGER committed
116
    # GO term id format: 'GO:0000000'
Hervé  MENAGER's avatar
Hervé MENAGER committed
117
118
    description = models.CharField('description', max_length=500)

119
    def autofill(self):
Hervé  MENAGER's avatar
Hervé MENAGER committed
120
121
122
123
        """
        fetch information from external services
        (EBI OLS)
        """
124
125
126
        info = get_go_info(self.go_id)
        self.description = info['label']

127
128
129
130
    @property
    def name(self):
        return self.go_id + ' ' + self.description

131
132
133
    def __str__(self):
        return self.description

Hervé  MENAGER's avatar
Hervé MENAGER committed
134

135
class Protein(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
136
137
138
139
    """
    Protein information (from Uniprot) 
    and the corresponding human-readable name
    """
Hervé  MENAGER's avatar
Hervé MENAGER committed
140
    uniprot_id = models.CharField('Uniprot ID', unique=True, max_length=10)
Hervé  MENAGER's avatar
Hervé MENAGER committed
141
142
    recommended_name_long = models.CharField(
        'Uniprot Recommended Name (long)', max_length=75)
Hervé  MENAGER's avatar
Hervé MENAGER committed
143
144
145
    short_name = models.CharField('Short name', max_length=50)
    gene_name = models.CharField('Gene name', unique=True, max_length=30)
    entry_name = models.CharField('Entry name', max_length=30)
146
    organism = models.ForeignKey('Taxonomy', models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
147
148
    molecular_functions = models.ManyToManyField(MolecularFunction)

149
    def autofill(self):
Hervé  MENAGER's avatar
Hervé MENAGER committed
150
151
152
153
        """
        fetch information from external services
        (Uniprot) and create Taxonomy/Molecular Functions if needed
        """
154
        info = get_uniprot_info(self.uniprot_id)
155
        self.recommended_name_long = info['recommended_name']
156
157
        self.gene_name = info['gene']
        self.entry_name = info['entry_name']
158
        self.short_name = info['short_name']
159
160
161
162
163
        try:
            taxonomy = Taxonomy.objects.get(taxonomy_id=info['organism'])
        except Taxonomy.DoesNotExist:
            taxonomy = Taxonomy()
            taxonomy.taxonomy_id = info['organism']
164
            taxonomy.save(autofill=True)
165
        self.organism = taxonomy
166
        super(Protein, self).save()
167
168
169
170
171
172
        for go_id in info['molecular_functions']:
            try:
                mol_function = MolecularFunction.objects.get(go_id=go_id)
            except MolecularFunction.DoesNotExist:
                mol_function = MolecularFunction()
                mol_function.go_id = go_id
173
                mol_function.save(autofill=True)
174
            self.molecular_functions.add(mol_function)
175

176
177
178
    def __str__(self):
        return '{} ({})'.format(self.uniprot_id, self.recommended_name_long)

Hervé  MENAGER's avatar
Hervé MENAGER committed
179

180
class Domain(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
181
182
183
    """
    Domain (i.e. Protein domain) information (from PFAM) 
    """
Hervé  MENAGER's avatar
Hervé MENAGER committed
184
185
    pfam_acc = models.CharField('Pfam Accession', max_length=10, unique=True)
    pfam_id = models.CharField('Pfam Family Identifier', max_length=20)
Hervé  MENAGER's avatar
Hervé MENAGER committed
186
    pfam_description = models.CharField('Pfam Description', max_length=100)
Hervé  MENAGER's avatar
Hervé MENAGER committed
187
    domain_family = models.CharField('Domain family', max_length=25)
Hervé  MENAGER's avatar
Hervé MENAGER committed
188
189
    # TODO: what is this field? check database
    # contents
190

191
    def autofill(self):
Hervé  MENAGER's avatar
Hervé MENAGER committed
192
193
194
195
        """
        fetch information from external services
        (PFAM)
        """
196
197
198
        info = get_pfam_info(self.pfam_acc)
        self.pfam_id = info['id']
        self.pfam_description = info['description']
Hervé  MENAGER's avatar
Hervé MENAGER committed
199

200
201
202
203
    @property
    def name(self):
        return self.pfam_id

204
205
206
    def __str__(self):
        return '{} ({}-{})'.format(self.pfam_acc, self.pfam_id, self.pfam_description)

Hervé  MENAGER's avatar
Hervé MENAGER committed
207

208
class ProteinDomainComplex(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
209
210
211
    """
    Protein-Domain association 
    """
212
213
    protein = models.ForeignKey('Protein', models.CASCADE)
    domain = models.ForeignKey('Domain', models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
214
215
216
    ppc_copy_nb = models.IntegerField(
        'Number of copies of the protein in the complex')

Hervé  MENAGER's avatar
Hervé MENAGER committed
217
218
    class Meta:
        verbose_name_plural = "complexes"
219

220
221
222
    def __str__(self):
        return '{}-{}'.format(self.protein_id, self.domain_id)

223
224
    def name(self):
        return self.protein.short_name
Hervé  MENAGER's avatar
Hervé MENAGER committed
225

226

227
class ProteinDomainBoundComplex(ProteinDomainComplex):
Hervé  MENAGER's avatar
Hervé MENAGER committed
228
229
230
    """
    Protein-Domain association with a "bound complex" role
    """
Hervé  MENAGER's avatar
Hervé MENAGER committed
231
232
233
    ppp_copy_nb_per_p = models.IntegerField(
        'Number of copies of the protein in the pocket')

234
235
    class Meta:
        verbose_name_plural = "bound complexes"
Hervé  MENAGER's avatar
Hervé MENAGER committed
236
237


238
class ProteinDomainPartnerComplex(ProteinDomainComplex):
Hervé  MENAGER's avatar
Hervé MENAGER committed
239
240
241
    """
    Protein-Domain association with a "partner complex" role
    """
Hervé  MENAGER's avatar
Hervé MENAGER committed
242

243
244
    class Meta:
        verbose_name_plural = "partner complexes"
Hervé  MENAGER's avatar
Hervé MENAGER committed
245

Hervé  MENAGER's avatar
Hervé MENAGER committed
246

247
class Symmetry(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
248
249
250
    """
    Symmetry of a PPI
    """
251
252
    code = models.CharField('Symmetry code', max_length=2)
    description = models.CharField('Description', max_length=300)
Hervé  MENAGER's avatar
Hervé MENAGER committed
253

254
255
256
    class Meta:
        verbose_name_plural = "symmetries"

257
258
259
    def __str__(self):
        return '{} ({})'.format(self.code, self.description)

260
261

class Disease(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
262
    name = models.CharField('Disease', max_length=30, unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
263
    # is there any database/nomenclature for diseases?
264
265
266
267

    def __str__(self):
        return self.name

Hervé  MENAGER's avatar
Hervé MENAGER committed
268
class PpiFamily(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
269
270
271
    """
    PPI Family
    """
Hervé  MENAGER's avatar
Hervé MENAGER committed
272
273
    name = models.CharField('Name', max_length=30, unique=True)

274
275
276
    class Meta:
        verbose_name_plural = "PPI Families"

Hervé  MENAGER's avatar
Hervé MENAGER committed
277
278
    def __str__(self):
        return self.name
Hervé  MENAGER's avatar
Hervé MENAGER committed
279

Hervé  MENAGER's avatar
Hervé MENAGER committed
280

281
class Ppi(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
282
283
284
    """
    PPI
    """
285
    pdb_id = models.CharField('PDB ID', max_length=4, null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
286
287
    pockets_nb = models.IntegerField(
        'Total number of pockets in the complex', default=1)
288
    symmetry = models.ForeignKey(Symmetry, models.CASCADE)
289
    diseases = models.ManyToManyField(Disease)
290
    family = models.ForeignKey(PpiFamily, models.CASCADE, null=True, blank=True)
291
    name = models.TextField('PPI name', null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
292

293
    def __str__(self):
294
        return 'PPI #{} on {}'.format(self.id, self.name)
295

296
297
298
    def autofill(self):
        # name is denormalized and stored in the database to reduce SQL queries in query mode
        self.name = self.compute_name_from_protein_names()
299
300
301
302
303

    def get_ppi_bound_complexes(self):
        """
        return bound ppi complexes belonging to this ppi
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
304
        # this is the less efficient query ever seen, FIXME
305
306
        return PpiComplex.objects.filter(ppi=self, complex__in=ProteinDomainBoundComplex.objects.all())

307
    def compute_name_from_protein_names(self):
Hervé  MENAGER's avatar
Hervé MENAGER committed
308
        all_protein_names = set(
309
            [ppi_complex.complex.protein.short_name for ppi_complex in self.ppicomplex_set.all()])
Hervé  MENAGER's avatar
Hervé MENAGER committed
310
311
        bound_protein_names = set(
            [ppi_complex.complex.protein.short_name for ppi_complex in self.get_ppi_bound_complexes()])
312
313
314
315
        partner_protein_names = all_protein_names - bound_protein_names
        bound_str = ','.join(bound_protein_names)
        partner_str = ','.join(partner_protein_names)
        name = bound_str
Hervé  MENAGER's avatar
Hervé MENAGER committed
316
        if partner_str != '':
317
318
            name += ' / ' + partner_str
        return name
319

Hervé  MENAGER's avatar
Hervé MENAGER committed
320

Hervé  MENAGER's avatar
Hervé MENAGER committed
321
class PpiComplex(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
322
323
324
    """
    PPI Complex
    """
325
326
    ppi = models.ForeignKey(Ppi, models.CASCADE)
    complex = models.ForeignKey(ProteinDomainComplex, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
327
328
    cc_nb = models.IntegerField(
        'Number of copies of the complex in the PPI', default=1)
Hervé  MENAGER's avatar
Hervé MENAGER committed
329
330
331
332

    class Meta:
        verbose_name_plural = "Ppi complexes"

333
334
335
    def __str__(self):
        return 'PPI {}, Complex {} ({})'.format(self.ppi, self.complex, self.cc_nb)

Hervé  MENAGER's avatar
Hervé MENAGER committed
336

337
class CompoundManager(models.Manager):
Hervé  MENAGER's avatar
Hervé MENAGER committed
338
339
340
341
    """
    CompoundManager adds automatically a number of annotations to the results
    of the database query, used for filters and compound card
    """
342
343
344

    def get_queryset(self):
        qs = super().get_queryset()
345
        # with number of publications
346
        qs = qs.annotate(pubs=Count('refcompoundbiblio', distinct=True))
347
        # with best activity
348
        qs = qs.annotate(best_activity=Max('compoundactivityresult__activity')) 
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
        # with LE
        qs = qs.annotate(le=Cast(1.37 * Max('compoundactivityresult__activity') / F('nb_atom_non_h'), FloatField()))
        # with LLE
        qs = qs.annotate(lle=Cast(Max('compoundactivityresult__activity') - F('a_log_p'), FloatField()))
        # Lipinsky MW (<=500)
        qs = qs.annotate(lipinsky_mw=Case(When(molecular_weight__lte=500, then=True), default=False, output_field=BooleanField()))
        # Lipinsky hba (<=10)
        qs = qs.annotate(lipinsky_hba=Case(When(nb_acceptor_h__lte=10, then=True), default=False, output_field=BooleanField()))
        # Lipinsky hbd (<5)
        qs = qs.annotate(lipinsky_hbd=Case(When(nb_donor_h__lte=5, then=True), default=False, output_field=BooleanField()))
        # Lipinsky a_log_p (<5)
        qs = qs.annotate(lipinsky_a_log_p=Case(When(a_log_p__lte=5, then=True), default=False, output_field=BooleanField()))
        # Lipinsky global
        qs = qs.annotate(lipinsky_score=Cast(F('lipinsky_mw'), IntegerField())+Cast(F('lipinsky_hba'), IntegerField())+ \
            Cast(F('lipinsky_hbd'), IntegerField()) + Cast(F('lipinsky_a_log_p'), IntegerField()))
        qs = qs.annotate(lipinsky=Case(When(lipinsky_score__gte=3, then=True), default=False, output_field=BooleanField()))
        # Veber hba_hbd (<=12)
        qs = qs.annotate(hba_hbd=F('nb_acceptor_h')+F('nb_donor_h'))
        qs = qs.annotate(veber_hba_hbd=Case(When(hba_hbd__lte=12, then=True), default=False, output_field=BooleanField()))
        # Veber TPSA (<=140)
        qs = qs.annotate(veber_tpsa=Case(When(tpsa__lte=140, then=True), default=False, output_field=BooleanField()))
        # Veber Rotatable Bonds (<=10)
        qs = qs.annotate(veber_rb=Case(When(nb_rotatable_bonds__lte=10, then=True), default=False, output_field=BooleanField()))
        # Veber global (Rotatable bonds and (hba_hbd or tpsa))
373
374
        #qs = qs.annotate(veber=F('veber_rb').bitand(F('veber_hba_hbd').bitor(F('veber_tpsa'))))
        qs = qs.annotate(veber=Case(When(Q(Q(nb_rotatable_bonds__lte=10) & (Q(hba_hbd__lte=12) | Q(tpsa__lte=140))), then=True), default=False, output_field=BooleanField()))
375
376
377
378
379
        # Pfizer AlogP (<=3)
        qs = qs.annotate(pfizer_a_log_p=Case(When(a_log_p__lte=3, then=True), default=False, output_field=BooleanField()))
        # Pfizer TPSA (>=75)
        qs = qs.annotate(pfizer_tpsa=Case(When(tpsa__gte=75, then=True), default=False, output_field=BooleanField()))
        # Pfizer global (AlogP and TPSA)
380
381
        #qs = qs.annotate(pfizer=F('pfizer_a_log_p').bitand(F('pfizer_tpsa')))
        qs = qs.annotate(pfizer=Case(When(Q(Q(a_log_p__lte=3) & Q(tpsa__gte=75)), then=True), default=False, output_field=BooleanField()))
382
        # PDB ligand available
383
        qs = qs.annotate(pdb_ligand_av=Cast(Max(Case(When(compoundaction__ligand_id__isnull=False, then=1), default=0, output_field=IntegerField())), BooleanField()))
384
385
386
387
388
389
        # inhibition role
        qs = qs.annotate(inhibition_role=Case(When(compoundactivityresult__modulation_type='I', then=True), default=False, output_field=BooleanField()))
        # binding role
        qs = qs.annotate(binding_role=Case(When(compoundactivityresult__modulation_type='B', then=True), default=False, output_field=BooleanField()))
        # stabilisation role
        qs = qs.annotate(stabilisation_role=Case(When(compoundactivityresult__modulation_type='S', then=True), default=False, output_field=BooleanField()))
390
        # cellular tests performed
391
        qs = qs.annotate(celltest_av=Cast(Max(Case(When(compoundactivityresult__test_activity_description__test_type='CELL', then=1), default=0, output_field=IntegerField())), BooleanField()))
392
        # inhibition tests performed
393
        qs = qs.annotate(inhitest_av=Cast(Max(Case(When(compoundactivityresult__test_activity_description__test_modulation_type='I', then=1), default=0, output_field=IntegerField())), BooleanField()))
394
        # stabilisation tests performed
395
        qs = qs.annotate(stabtest_av=Cast(Max(Case(When(compoundactivityresult__test_activity_description__test_modulation_type='S', then=1), default=0, output_field=IntegerField())), BooleanField()))
396
        # binding tests performed
397
        qs = qs.annotate(bindtest_av=Cast(Max(Case(When(compoundactivityresult__test_activity_description__test_modulation_type='B', then=1), default=0, output_field=IntegerField())), BooleanField()))
398
        # pharmacokinetic tests performed
399
        qs = qs.annotate(pktest_av=Cast(Max(Case(When(refcompoundbiblio__bibliography__pharmacokinetic=True, then=1), default=0, output_field=IntegerField())), BooleanField()))
400
        # cytotoxicity tests performedudy
401
        qs = qs.annotate(cytoxtest_av=Cast(Max(Case(When(refcompoundbiblio__bibliography__cytotox=True, then=1), default=0, output_field=IntegerField())), BooleanField()))
402
        # in silico st performed
403
        qs = qs.annotate(insilico_av=Cast(Max(Case(When(refcompoundbiblio__bibliography__in_silico=True, then=1), default=0, output_field=IntegerField())), BooleanField()))
404
405
        return qs

406
class Compound(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
407
408
409
    """
    Chemical compound
    """
410
    objects = CompoundManager() 
411
412
    canonical_smile = models.TextField(
        'Canonical Smile', unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
413
414
415
416
417
    is_macrocycle = models.BooleanField('Contains one or more macrocycles')
    aromatic_ratio = models.DecimalField(
        'Aromatic ratio', max_digits=3, decimal_places=2)
    balaban_index = models.DecimalField(
        'Balaban index', max_digits=3, decimal_places=2)
418
    fsp3 = models.DecimalField('Fsp3', max_digits=3, decimal_places=2)
Hervé  MENAGER's avatar
Hervé MENAGER committed
419
420
421
422
423
424
425
426
427
428
    gc_molar_refractivity = models.DecimalField(
        'GC Molar Refractivity', max_digits=5, decimal_places=2)
    log_d = models.DecimalField(
        'LogD (Partition coefficient octanol-1/water, with pKa information)', max_digits=4, decimal_places=2)
    a_log_p = models.DecimalField(
        'ALogP (Partition coefficient octanol-1/water)', max_digits=4, decimal_places=2)
    mean_atom_vol_vdw = models.DecimalField(
        'Mean atom volume computed with VdW radii', max_digits=4, decimal_places=2)
    molecular_weight = models.DecimalField(
        'Molecular weight', max_digits=6, decimal_places=2)
429
430
431
432
    nb_acceptor_h = models.IntegerField('Number of hydrogen bond acceptors')
    nb_aliphatic_amines = models.IntegerField('Number of aliphatics amines')
    nb_aromatic_bonds = models.IntegerField('Number of aromatic bonds')
    nb_aromatic_ether = models.IntegerField('Number of aromatic ethers')
Hervé  MENAGER's avatar
Hervé MENAGER committed
433
434
435
436
    nb_aromatic_sssr = models.IntegerField(
        'Number of aromatic Smallest Set of System Rings (SSSR)')
    nb_atom = models.IntegerField('Number of atoms')
    nb_atom_non_h = models.IntegerField('Number of non hydrogen atoms')
437
438
    nb_benzene_like_rings = models.IntegerField('Number of benzene-like rings')
    nb_bonds = models.IntegerField('Number of bonds')
Hervé  MENAGER's avatar
Hervé MENAGER committed
439
440
441
442
443
444
445
446
447
448
449
450
451
452
    nb_bonds_non_h = models.IntegerField(
        'Number of bonds not involving a hydrogen')
    nb_br = models.IntegerField('Number of Bromine atoms')
    nb_c = models.IntegerField('Number of Carbon atoms')
    nb_chiral_centers = models.IntegerField('Number of chiral centers')
    nb_circuits = models.IntegerField('Number of circuits')
    nb_cl = models.IntegerField('Number of Chlorine atoms')
    nb_csp2 = models.IntegerField('Number of sp2-hybridized carbon atoms')
    nb_csp3 = models.IntegerField('Number of sp3-hybridized carbon atoms')
    nb_donor_h = models.IntegerField('Number of hydrogen bond donors')
    nb_double_bonds = models.IntegerField('Number of double bonds')
    nb_f = models.IntegerField('Number of fluorine atoms')
    nb_i = models.IntegerField('Number of iodine atoms')
    nb_multiple_bonds = models.IntegerField('Number of multiple bonds')
453
    nb_n = models.IntegerField('Number of nitrogen atoms')
Hervé  MENAGER's avatar
Hervé MENAGER committed
454
455
456
    nb_o = models.IntegerField('Number of oxygen atoms')
    nb_rings = models.IntegerField('Number of rings')
    nb_rotatable_bonds = models.IntegerField('Number of rotatable bonds')
457
458
    inchi = models.TextField('InChi')
    inchikey = models.TextField('InChiKey')
Hervé  MENAGER's avatar
Hervé MENAGER committed
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
    randic_index = models.DecimalField(
        'Randic index', max_digits=4, decimal_places=2)
    rdf070m = models.DecimalField(
        'RDF070m, radial distribution function weighted by the atomic masses at 7Å', max_digits=5, decimal_places=2)
    rotatable_bond_fraction = models.DecimalField(
        'Fraction of rotatable bonds', max_digits=3, decimal_places=2)
    sum_atom_polar = models.DecimalField(
        'Sum of atomic polarizabilities', max_digits=5, decimal_places=2)
    sum_atom_vol_vdw = models.DecimalField(
        'Sum of atom volumes computed with VdW radii', max_digits=6, decimal_places=2)
    tpsa = models.DecimalField(
        'Topological Polar Surface Area (TPSA)', max_digits=5, decimal_places=2)
    ui = models.DecimalField(
        'Unsaturation index', max_digits=4, decimal_places=2)
    wiener_index = models.IntegerField('Wiener index')
    common_name = models.CharField(
        'Common name', unique=True, max_length=20, blank=True, null=True)
    pubchem_id = models.CharField(
        'Pubchem ID', max_length=10, blank=True, null=True)
    chemspider_id = models.CharField(
        'Chemspider ID', unique=True, max_length=10, blank=True, null=True)
    chembl_id = models.CharField(
        'Chembl ID', max_length=30, blank=True, null=True)
    iupac_name = models.CharField(
        'IUPAC name', max_length=255, blank=True, null=True)
484

485
486
487
    class Meta:
       ordering = ['id']

488
489
490
    def compute_drugbank_compound_similarity(self):
        """ compute Tanimoto similarity to existing DrugBank compounds """
        self.save()
491
        # fingerprints to compute drugbank similarities are in settings module, default FP2
492
        fingerprinter = FingerPrinter(getattr(settings, "DRUGBANK_FINGERPRINTS", "FP2"))
493
494
495
496
497
498
499
500
501
        #1. compute tanimoto for SMILES query vs all compounds
        smiles_dict = {c.id:c.canonical_smiles for c in DrugBankCompound.objects.all()}
        tanimoto_dict = fingerprinter.tanimoto_smiles(self.canonical_smile, smiles_dict)
        tanimoto_dict = dict(sorted(tanimoto_dict.items(), key=operator.itemgetter(1), reverse=True)[:15])
        dbcts = []
        for id_, tanimoto in tanimoto_dict.items():
            dbcts.append(DrugbankCompoundTanimoto(compound=self, drugbank_compound=DrugBankCompound.objects.get(id=id_), tanimoto=tanimoto))
        DrugbankCompoundTanimoto.objects.bulk_create(dbcts)

Hervé  MENAGER's avatar
Hervé MENAGER committed
502
503
504
    @property
    def biblio_refs(self):
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
505
        return all RefCompoundBiblio related to this compound
Hervé  MENAGER's avatar
Hervé MENAGER committed
506
507
        """
        return RefCompoundBiblio.objects.filter(compound=self)
508

509
510
511
512
513
514
515
516
517
518
519
520
521
    @property
    def pfam_ids(self):
        """
        return all PFAM ids for the domain of the proteins of the bound
        complexes in the PPIs this compound has an action on
        """
        pfam_ids = set()
        for ca in self.compoundaction_set.all():
            ca.get_complexes()
            for bound_complex in ca.ppi.get_ppi_bound_complexes():
                pfam_ids.add(bound_complex.complex.domain.pfam_id)
        return pfam_ids

522
    @property
Hervé  MENAGER's avatar
Hervé MENAGER committed
523
    def compound_action_ligand_ids(self):
524
525
526
        """
        return all PDB codes of the corresponding compound actions
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
527
        ligand_ids = set()
528
        for ca in self.compoundaction_set.all():
Hervé  MENAGER's avatar
Hervé MENAGER committed
529
530
            ligand_ids.add(ca.ligand_id)
        return ligand_ids
531

532
533
    @property
    def best_pXC50_activity(self):
534
        return self.compoundactivityresult_set.aggregate(Max('activity'))['activity__max']
535
536
537
538
539
540

    @property
    def best_pXC50_compound_activity_result(self):
        best_pXC50_activity = self.best_pXC50_activity
        if best_pXC50_activity is None:
            return None
541
        return self.compoundactivityresult_set.filter(activity=best_pXC50_activity)[0]
542

543
    @property
544
    def best_pXC50_activity_ppi_name(self):
545
        """
546
        Name of the PPI corresponding to the best PXC50 activity
547
        """
548
549
550
        best_activity_car = self.best_pXC50_compound_activity_result
        if best_activity_car is None:
            return None
551
552
553
        ppi_name = best_activity_car.test_activity_description.ppi.name
        return ppi_name

554
555
556
557
558
559
560
561
562
563
564
    @property
    def best_pXC50_activity_ppi_family(self):
        """
        Family of the PPI corresponding to the best PXC50 activity
        """
        best_activity_car = self.best_pXC50_compound_activity_result
        if best_activity_car is None:
            return None
        ppi_family = best_activity_car.test_activity_description.ppi.family.name
        return ppi_family

565
566
567
568
569
570
571
572
573
574
575
576
577
578
    @property
    def bioch_tests_count(self):
        """
        return the number of associated biochemical tests
        """
        return self.compoundactivityresult_set.all().filter(test_activity_description__test_type='BIOCH').count()

    @property
    def cell_tests_count(self):
        """
        return the number of associated cell tests
        """
        return self.compoundactivityresult_set.all().filter(test_activity_description__test_type='CELL').count()

579
580
581
582
583
584
    @property
    def families(self):
        """
        return the all PPI families for PPIs involved in the compound activity of the compound
        """
        return list(set([ca.ppi.family for ca in self.compoundaction_set.all()]))
585
586
587
588
589
    
    @property
    def sorted_similar_drugbank_compounds(self):
        return self.drugbankcompoundtanimoto_set.order_by('-tanimoto')
        
590
591
592
    def autofill(self):
        # compute InChi and InChiKey
        self.inchi = smi2inchi(self.canonical_smile)
593
        self.inchikey = smi2inchikey(self.canonical_smile)
594
        self.compute_drugbank_compound_similarity()
595

596
597
598
    def __str__(self):
        return 'Compound #{}'.format(self.id)

599
600
601
class CompoundTanimoto(models.Model):
    canonical_smiles = models.TextField(
        'Canonical Smile')
602
    fingerprint = models.TextField('Fingerprint')
603
604
605
606
607
608
    compound = models.ForeignKey(Compound, models.CASCADE)
    tanimoto = models.DecimalField(
        'Tanimoto value', max_digits=5, decimal_places=4)

    class Meta:
        unique_together = (
609
            ('canonical_smiles', 'fingerprint', 'compound'))
610

611
def create_tanimoto(smiles_query, fingerprint):
Hervé  MENAGER's avatar
Hervé MENAGER committed
612
613
614
615
    """
    Compute the Tanimoto similarity between a given SMILES and the compounds
    then insert the results in CompoundTanimoto
    """
616
    if CompoundTanimoto.objects.filter(canonical_smiles=smiles_query, fingerprint=fingerprint).count()==0:
617
        smiles_dict = {c.id:c.canonical_smile for c in Compound.objects.all()}
618
        fingerprinter = FingerPrinter(fingerprint)
619
620
621
622
623
        #1. compute tanimoto for SMILES query vs all compounds
        tanimoto_dict = fingerprinter.tanimoto_smiles(smiles_query, smiles_dict)
        #2. insert results in a table with three fields: SMILES query, compound id, tanimoto index
        cts = []
        for id_, smiles in smiles_dict.items():
624
            cts.append(CompoundTanimoto(canonical_smiles=smiles_query, fingerprint=fingerprint, compound=Compound.objects.get(id=id_), tanimoto=tanimoto_dict[id_]))
625
        CompoundTanimoto.objects.bulk_create(cts)
626

627
class PcaBiplotData(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
628
629
630
631
    """
    PCA biplot data
    the table contains all the data as one JSON text in one row
    """
632
633
    pca_biplot_data = models.TextField(
        'PCA biplot JSON data', blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
634

635

636
class LeLleBiplotData(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
637
638
639
640
    """
    LE-LLE biplot data
    the table contains all the data as one JSON text in one row
    """
641
642
    le_lle_biplot_data = models.TextField(
        'LE-LLE biplot JSON data', blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
643

644

645
class CellLine(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
646
647
648
    """
    Cell lines
    """
649
    name = models.CharField('Name', max_length=50, unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
650

651
652
653
    def __str__(self):
        return self.name

Hervé  MENAGER's avatar
Hervé MENAGER committed
654

655
class TestActivityDescription(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
656
657
658
    """
    Activity test descriptions
    """
659
660
661
662
663
664
665
666
667
    TEST_TYPES = (
        ('BIOCH', 'Biochemical assay'),
        ('CELL', 'Cellular assay')
    )
    TEST_MODULATION_TYPES = (
        ('B', 'Binding'),
        ('I', 'Inhibition'),
        ('S', 'Stabilization')
    )
668
669
670
671
    PROTEIN_BOUND_CONSTRUCTS = (
        ('F', 'Full length'),
        ('U', 'Unspecified')
    )
672
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
673
674
    protein_domain_bound_complex = models.ForeignKey(
        ProteinDomainBoundComplex, models.CASCADE)
675
    ppi = models.ForeignKey(Ppi, models.CASCADE, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
676
    test_name = models.CharField('Test name', max_length=100)
677
    is_primary = models.BooleanField('Is primary')
Hervé  MENAGER's avatar
Hervé MENAGER committed
678
679
    protein_bound_construct = models.CharField(
        'Protein bound construct', max_length=5, choices=PROTEIN_BOUND_CONSTRUCTS, blank=True, null=True)
680
    test_type = models.CharField('Test type', max_length=5, choices=TEST_TYPES)
Hervé  MENAGER's avatar
Hervé MENAGER committed
681
682
683
684
    test_modulation_type = models.CharField(
        'Test modulation type', max_length=1, choices=TEST_MODULATION_TYPES)
    nb_active_compounds = models.IntegerField(
        'Total number of active compounds')
Hervé  MENAGER's avatar
Hervé MENAGER committed
685
686
    cell_line = models.ForeignKey(
        CellLine, models.CASCADE, blank=True, null=True)
687

Hervé  MENAGER's avatar
Hervé MENAGER committed
688
    def get_complexes(self):
689
690
691
692
        """
        get the complexes tested for this PPI
        depends on the modulation type
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
693
        if self.test_modulation_type == 'I':
694
            return self.ppi.ppicomplex_set.all()
695
696
        else:
            return self.ppi.get_ppi_bound_complexes()
697

698
699
700
    @property
    def protein_domain_partner_complex(self):
        for ppic in self.ppi.ppicomplex_set.all():
Hervé  MENAGER's avatar
Hervé MENAGER committed
701
            if hasattr(ppic.complex, 'proteindomainpartnercomplex'):
702
703
704
                return ppic.complex.proteindomainpartnercomplex
        return None

705
706
707
708
709
710
711
    @property
    def name(self):
        return self.test_name

    def __str__(self):
        return self.get_test_type_display()

Hervé  MENAGER's avatar
Hervé MENAGER committed
712

713
class CompoundActivityResult(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
714
715
716
    """
    Activity test results on a compound
    """
717
718
719
720
    MODULATION_TYPES = (
        ('I', 'Inhibition'),
        ('S', 'Stabilization')
    )
Hervé  MENAGER's avatar
Hervé MENAGER committed
721
    ACTIVITY_TYPES = (
Hervé  MENAGER's avatar
Hervé MENAGER committed
722
723
724
725
        ('pIC50', 'pIC50 (half maximal inhibitory concentration, -log10)'),
        ('pEC50', 'pEC50 (half maximal effective concentration, -log10)'),
        ('pKd', 'pKd (dissociation constant, -log10)'),
        ('pKi', 'pKi (inhibition constant, -log10)'),
Hervé  MENAGER's avatar
Hervé MENAGER committed
726
    )
727
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
728
729
    test_activity_description = models.ForeignKey(
        TestActivityDescription, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
730
731
732
733
    activity_type = models.CharField(
        'Activity type', max_length=5, choices=ACTIVITY_TYPES)
    activity = models.DecimalField(
        'Activity', max_digits=12, decimal_places=10)
734
    inhibition_percentage = models.DecimalField(
735
        'Inhibition percentage', max_digits=3, decimal_places=0, null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
736
737
    modulation_type = models.CharField(
        'Modulation type', max_length=1, choices=MODULATION_TYPES)
Hervé  MENAGER's avatar
Hervé MENAGER committed
738
739

    class Meta:
Hervé  MENAGER's avatar
Hervé MENAGER committed
740
741
742
        unique_together = (
            ('compound', 'test_activity_description', 'activity_type'),)

743
744
    def __str__(self):
        return 'Compound activity result for {} test {} on {}'.format(self.activity_type, self.test_activity_description.id, self.compound.id)
Hervé  MENAGER's avatar
Hervé MENAGER committed
745

746
747
748
    def is_best(self):
        return self.compound.best_pXC50_compound_activity_result.id == self.id

Hervé  MENAGER's avatar
Hervé MENAGER committed
749

750
class TestCytotoxDescription(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
751
752
753
    """
    Cytotoxicity test descriptions
    """
754
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
755
    test_name = models.CharField('Cytotoxicity test name', max_length=100)
756
    cell_line = models.ForeignKey(CellLine, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
757
758
759
    compound_concentration = models.DecimalField(
        'Compound concentration in μM', max_digits=7, decimal_places=3, blank=True, null=True)

Hervé  MENAGER's avatar
Hervé MENAGER committed
760

761
class CompoundCytotoxicityResult(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
762
763
764
    """
    Cytotoxicity test results on a compound
    """
765
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
766
767
    test_cytotoxicity_description = models.ForeignKey(
        TestCytotoxDescription, models.CASCADE)
768
    toxicity = models.BooleanField('Toxicity', default=False)
Hervé  MENAGER's avatar
Hervé MENAGER committed
769
770

    class Meta:
771
        unique_together = (('compound', 'test_cytotoxicity_description'),)
772

773
774
775
    def __str__(self):
        return 'Compound cytotoxicity result for test {} on {}'.format(self.test_cytotoxicity_description.id, self.compound.id)

Hervé  MENAGER's avatar
Hervé MENAGER committed
776

777
class TestPKDescription(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
778
779
780
    """
    Pharmacokinetic test descriptions
    """
781
782
783
784
785
786
    ADMINISTRATION_MODES = (
        ('IV', ''),
        ('PO', ''),
        ('IP', ''),
        ('SL', 'SL')
    )
787
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
788
    test_name = models.CharField('Pharmacokinetic test name', max_length=100)
789
    organism = models.ForeignKey(Taxonomy, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
790
791
    administration_mode = models.CharField(
        'Administration mode', max_length=2, choices=ADMINISTRATION_MODES, blank=True, null=True)
792
793
    concentration = models.DecimalField(
        'Concentration in mg/l', max_digits=7, decimal_places=3, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
794
    dose = models.DecimalField(
795
        'Dose in mg/kg', max_digits=9, decimal_places=4, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
796
797
798
    dose_interval = models.IntegerField(
        'Dose interval, in hours', blank=True, null=True)

799
800

class CompoundPKResult(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
801
802
803
    """
    Pharmacokinetic test results on a compound
    """
804
805
    compound = models.ForeignKey(Compound, models.CASCADE)
    test_pk_description = models.ForeignKey(TestPKDescription, models.CASCADE)
806
    tolerated = models.NullBooleanField('Tolerated', null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
807
808
809
810
    auc = models.IntegerField(
        'Area under curve (ng.mL-1.hr)', blank=True, null=True)
    clearance = models.DecimalField(
        'Clearance (mL/hr)', max_digits=7, decimal_places=3, blank=True, null=True)
811
    c_max = models.DecimalField(
Hervé  MENAGER's avatar
Hervé MENAGER committed
812
813
814
815
816
817
818
        'Maximal concentration (ng/mL)', max_digits=7, decimal_places=3, blank=True, null=True)
    oral_bioavailability = models.IntegerField(
        'Oral Bioavailability (%F)', blank=True, null=True)
    t_demi = models.IntegerField('t½', blank=True, null=True)
    t_max = models.IntegerField('tmax', blank=True, null=True)
    voldistribution = models.DecimalField(
        'Volume distribution (Vd)', max_digits=5, decimal_places=2, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
819
820

    class Meta:
821
        unique_together = (('compound', 'test_pk_description'),)
Hervé  MENAGER's avatar
Hervé MENAGER committed
822

823
824
825
    def __str__(self):
        return 'Compound PK result for test {} on {}'.format(self.test_pk_description.id, self.compound.id)

Hervé  MENAGER's avatar
Hervé MENAGER committed
826

827
class CompoundAction(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
828
829
830
    """
    Compound action
    """
831
832
    ACTIVATION_MODES = (
        ('O', 'Orthosteric'),
833
834
        ('A', 'Allosteric'),
        ('U', 'Unspecified')
835
    )
836
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
837
838
    activation_mode = models.CharField(
        'Activation mode', max_length=1, choices=ACTIVATION_MODES)
839
    ppi = models.ForeignKey(Ppi, models.CASCADE)
840
    ligand_id = models.CharField('PDB Ligand ID', max_length=3, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
841
842
    nb_copy_compounds = models.IntegerField(
        'Number of copies for the compound')
Hervé  MENAGER's avatar
Hervé MENAGER committed
843
844

    class Meta:
845
        unique_together = (('ppi', 'compound', 'activation_mode', 'ligand_id'),)
Hervé  MENAGER's avatar
Hervé MENAGER committed
846
847

    def get_complexes(self):
848
849
850
851
        """
        get the complexes involved in the compound action
        which are always the bound complexes
        """
852
        return self.ppi.get_ppi_bound_complexes()
Hervé  MENAGER's avatar
Hervé MENAGER committed
853

854
855
    def __str__(self):
        return 'Action of {} on {}'.format(self.compound, self.ppi)
Hervé  MENAGER's avatar
Hervé MENAGER committed
856

857
class RefCompoundBiblio(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
858
859
860
    """
    Compound-Bibliographic reference association
    """
861
862
    compound = models.ForeignKey(Compound, models.CASCADE)
    bibliography = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
863
864
    compound_name = models.CharField(
        'Compound name in the publication', max_length=50)
Hervé  MENAGER's avatar
Hervé MENAGER committed
865
866

    class Meta:
867
        unique_together = (('compound', 'bibliography'),)
868

869
class DrugBankCompound(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
870
871
872
    """
    Drugbank compound
    """
873
874
875
876
877
    id = models.TextField(
        'Drugbank ID', unique=True, primary_key=True)
    common_name = models.TextField('Common name')
    canonical_smiles = models.TextField(
        'Canonical SMILES')
878
879

class DrugbankCompoundTanimoto(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
880
881
882
    """
    Drugbank compound-compound tanimoto similarity
    """
883
884
885
    compound = models.ForeignKey(Compound, models.CASCADE)
    drugbank_compound = models.ForeignKey(DrugBankCompound, models.CASCADE)
    tanimoto = models.DecimalField(
886
        'Tanimoto value', max_digits=5, decimal_places=4)