models.py 32.7 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
43
44
    """
    Bibliography data table
    """
    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):
66
67
68
        if self.source == 'PM':
            info = get_pubmed_info(self.id_source)
        else:
69
            info = get_google_patent_info(self.id_source)
70
71
72
73
        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
74

Hervé  MENAGER's avatar
Hervé MENAGER committed
75
76
77
    class Meta:
        verbose_name_plural = "bibliographies"

78
79
    def __str__(self):
        return '{}, {}'.format(self.source, self.id_source)
80

Hervé  MENAGER's avatar
Hervé MENAGER committed
81

82
class Taxonomy(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
83
84
    taxonomy_id = models.DecimalField(
        'NCBI TaxID', unique=True, max_digits=9, decimal_places=0)
Hervé  MENAGER's avatar
Hervé MENAGER committed
85
    name = models.CharField('Organism name', max_length=200)
86

87
    def autofill(self):
88
89
90
        info = get_taxonomy_info(self.taxonomy_id)
        self.name = info['scientific_name']

91
92
93
    def __str__(self):
        return self.name

Hervé  MENAGER's avatar
Hervé MENAGER committed
94
95
    class Meta:
        verbose_name_plural = "taxonomies"
Hervé  MENAGER's avatar
Hervé MENAGER committed
96

Hervé  MENAGER's avatar
Hervé MENAGER committed
97

98
class MolecularFunction(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
99
    go_id = models.CharField('Gene Ontology ID', unique=True, max_length=10)
Hervé  MENAGER's avatar
Hervé MENAGER committed
100
    # GO term id format: 'GO:0000000'
Hervé  MENAGER's avatar
Hervé MENAGER committed
101
102
    description = models.CharField('description', max_length=500)

103
    def autofill(self):
104
105
106
        info = get_go_info(self.go_id)
        self.description = info['label']

107
108
109
110
    @property
    def name(self):
        return self.go_id + ' ' + self.description

111
112
113
    def __str__(self):
        return self.description

Hervé  MENAGER's avatar
Hervé MENAGER committed
114

115
class Protein(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
116
    uniprot_id = models.CharField('Uniprot ID', unique=True, max_length=10)
Hervé  MENAGER's avatar
Hervé MENAGER committed
117
118
    recommended_name_long = models.CharField(
        'Uniprot Recommended Name (long)', max_length=75)
Hervé  MENAGER's avatar
Hervé MENAGER committed
119
120
121
    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)
122
    organism = models.ForeignKey('Taxonomy', models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
123
124
    molecular_functions = models.ManyToManyField(MolecularFunction)

125
    def autofill(self):
126
        info = get_uniprot_info(self.uniprot_id)
127
        self.recommended_name_long = info['recommended_name']
128
129
        self.gene_name = info['gene']
        self.entry_name = info['entry_name']
130
        self.short_name = info['short_name']
131
132
133
134
135
        try:
            taxonomy = Taxonomy.objects.get(taxonomy_id=info['organism'])
        except Taxonomy.DoesNotExist:
            taxonomy = Taxonomy()
            taxonomy.taxonomy_id = info['organism']
136
            taxonomy.save(autofill=True)
137
        self.organism = taxonomy
138
        super(Protein, self).save()
139
140
141
142
143
144
        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
145
                mol_function.save(autofill=True)
146
            self.molecular_functions.add(mol_function)
147

148
149
150
    def __str__(self):
        return '{} ({})'.format(self.uniprot_id, self.recommended_name_long)

Hervé  MENAGER's avatar
Hervé MENAGER committed
151

152
class Domain(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
153
154
    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
155
    pfam_description = models.CharField('Pfam Description', max_length=100)
Hervé  MENAGER's avatar
Hervé MENAGER committed
156
    domain_family = models.CharField('Domain family', max_length=25)
Hervé  MENAGER's avatar
Hervé MENAGER committed
157
158
    # TODO: what is this field? check database
    # contents
159

160
    def autofill(self):
161
162
163
        info = get_pfam_info(self.pfam_acc)
        self.pfam_id = info['id']
        self.pfam_description = info['description']
Hervé  MENAGER's avatar
Hervé MENAGER committed
164

165
166
167
168
    @property
    def name(self):
        return self.pfam_id

169
170
171
    def __str__(self):
        return '{} ({}-{})'.format(self.pfam_acc, self.pfam_id, self.pfam_description)

Hervé  MENAGER's avatar
Hervé MENAGER committed
172

173
class ProteinDomainComplex(models.Model):
174
175
    protein = models.ForeignKey('Protein', models.CASCADE)
    domain = models.ForeignKey('Domain', models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
176
177
178
    ppc_copy_nb = models.IntegerField(
        'Number of copies of the protein in the complex')

Hervé  MENAGER's avatar
Hervé MENAGER committed
179
180
    class Meta:
        verbose_name_plural = "complexes"
181

182
183
184
    def __str__(self):
        return '{}-{}'.format(self.protein_id, self.domain_id)

185
186
    def name(self):
        return self.protein.short_name
Hervé  MENAGER's avatar
Hervé MENAGER committed
187

188

189
class ProteinDomainBoundComplex(ProteinDomainComplex):
Hervé  MENAGER's avatar
Hervé MENAGER committed
190
191
192
    ppp_copy_nb_per_p = models.IntegerField(
        'Number of copies of the protein in the pocket')

193
194
    class Meta:
        verbose_name_plural = "bound complexes"
Hervé  MENAGER's avatar
Hervé MENAGER committed
195
196


197
class ProteinDomainPartnerComplex(ProteinDomainComplex):
Hervé  MENAGER's avatar
Hervé MENAGER committed
198

199
200
    class Meta:
        verbose_name_plural = "partner complexes"
Hervé  MENAGER's avatar
Hervé MENAGER committed
201

Hervé  MENAGER's avatar
Hervé MENAGER committed
202

203
204
205
class Symmetry(models.Model):
    code = models.CharField('Symmetry code', max_length=2)
    description = models.CharField('Description', max_length=300)
Hervé  MENAGER's avatar
Hervé MENAGER committed
206

207
208
209
    class Meta:
        verbose_name_plural = "symmetries"

210
211
212
    def __str__(self):
        return '{} ({})'.format(self.code, self.description)

213
214

class Disease(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
215
    name = models.CharField('Disease', max_length=30, unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
216
    # is there any database/nomenclature for diseases?
217
218
219
220

    def __str__(self):
        return self.name

Hervé  MENAGER's avatar
Hervé MENAGER committed
221
222
223
class PpiFamily(models.Model):
    name = models.CharField('Name', max_length=30, unique=True)

224
225
226
    class Meta:
        verbose_name_plural = "PPI Families"

Hervé  MENAGER's avatar
Hervé MENAGER committed
227
228
    def __str__(self):
        return self.name
Hervé  MENAGER's avatar
Hervé MENAGER committed
229

Hervé  MENAGER's avatar
Hervé MENAGER committed
230

231
class Ppi(AutoFillableModel):
232
    pdb_id = models.CharField('PDB ID', max_length=4, null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
233
234
    pockets_nb = models.IntegerField(
        'Total number of pockets in the complex', default=1)
235
    symmetry = models.ForeignKey(Symmetry, models.CASCADE)
236
    diseases = models.ManyToManyField(Disease)
237
    family = models.ForeignKey(PpiFamily, models.CASCADE, null=True, blank=True)
238
    name = models.TextField('PPI name', null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
239

240
    def __str__(self):
241
        return 'PPI #{} on {}'.format(self.id, self.name)
242

243
244
245
    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()
246
247
248
249
250

    def get_ppi_bound_complexes(self):
        """
        return bound ppi complexes belonging to this ppi
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
251
        # this is the less efficient query ever seen, FIXME
252
253
        return PpiComplex.objects.filter(ppi=self, complex__in=ProteinDomainBoundComplex.objects.all())

254
    def compute_name_from_protein_names(self):
Hervé  MENAGER's avatar
Hervé MENAGER committed
255
        all_protein_names = set(
256
            [ppi_complex.complex.protein.short_name for ppi_complex in self.ppicomplex_set.all()])
Hervé  MENAGER's avatar
Hervé MENAGER committed
257
258
        bound_protein_names = set(
            [ppi_complex.complex.protein.short_name for ppi_complex in self.get_ppi_bound_complexes()])
259
260
261
262
        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
263
        if partner_str != '':
264
265
            name += ' / ' + partner_str
        return name
266

Hervé  MENAGER's avatar
Hervé MENAGER committed
267

Hervé  MENAGER's avatar
Hervé MENAGER committed
268
class PpiComplex(models.Model):
269
270
    ppi = models.ForeignKey(Ppi, models.CASCADE)
    complex = models.ForeignKey(ProteinDomainComplex, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
271
272
    cc_nb = models.IntegerField(
        'Number of copies of the complex in the PPI', default=1)
Hervé  MENAGER's avatar
Hervé MENAGER committed
273
274
275
276

    class Meta:
        verbose_name_plural = "Ppi complexes"

277
278
279
    def __str__(self):
        return 'PPI {}, Complex {} ({})'.format(self.ppi, self.complex, self.cc_nb)

Hervé  MENAGER's avatar
Hervé MENAGER committed
280

281
282
283
284
class CompoundManager(models.Manager):

    def get_queryset(self):
        qs = super().get_queryset()
285
        # with number of publications
286
        qs = qs.annotate(pubs=Count('refcompoundbiblio', distinct=True))
287
        # with best activity
288
        qs = qs.annotate(best_activity=Max('compoundactivityresult__activity')) 
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
        # 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))
313
314
        #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()))
315
316
317
318
319
        # 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)
320
321
        #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()))
322
        # PDB ligand available
323
        qs = qs.annotate(pdb_ligand_av=Cast(Max(Case(When(compoundaction__ligand_id__isnull=False, then=1), default=0, output_field=IntegerField())), BooleanField()))
324
325
326
327
328
329
        # 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()))
330
        # cellular tests performed
331
        qs = qs.annotate(celltest_av=Cast(Max(Case(When(compoundactivityresult__test_activity_description__test_type='CELL', then=1), default=0, output_field=IntegerField())), BooleanField()))
332
        # inhibition tests performed
333
        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()))
334
        # stabilisation tests performed
335
        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()))
336
        # binding tests performed
337
        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()))
338
        # pharmacokinetic tests performed
339
        qs = qs.annotate(pktest_av=Cast(Max(Case(When(refcompoundbiblio__bibliography__pharmacokinetic=True, then=1), default=0, output_field=IntegerField())), BooleanField()))
340
        # cytotoxicity tests performedudy
341
        qs = qs.annotate(cytoxtest_av=Cast(Max(Case(When(refcompoundbiblio__bibliography__cytotox=True, then=1), default=0, output_field=IntegerField())), BooleanField()))
342
        # in silico st performed
343
        qs = qs.annotate(insilico_av=Cast(Max(Case(When(refcompoundbiblio__bibliography__in_silico=True, then=1), default=0, output_field=IntegerField())), BooleanField()))
344
345
        return qs

346
class Compound(AutoFillableModel):
347
    objects = CompoundManager() 
348
349
    canonical_smile = models.TextField(
        'Canonical Smile', unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
350
351
352
353
354
    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)
355
    fsp3 = models.DecimalField('Fsp3', max_digits=3, decimal_places=2)
Hervé  MENAGER's avatar
Hervé MENAGER committed
356
357
358
359
360
361
362
363
364
365
    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)
366
367
368
369
    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
370
371
372
373
    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')
374
375
    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
376
377
378
379
380
381
382
383
384
385
386
387
388
389
    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')
390
    nb_n = models.IntegerField('Number of nitrogen atoms')
Hervé  MENAGER's avatar
Hervé MENAGER committed
391
392
393
    nb_o = models.IntegerField('Number of oxygen atoms')
    nb_rings = models.IntegerField('Number of rings')
    nb_rotatable_bonds = models.IntegerField('Number of rotatable bonds')
394
395
    inchi = models.TextField('InChi')
    inchikey = models.TextField('InChiKey')
Hervé  MENAGER's avatar
Hervé MENAGER committed
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
    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)
421

422
423
424
    class Meta:
       ordering = ['id']

425
426
427
    def compute_drugbank_compound_similarity(self):
        """ compute Tanimoto similarity to existing DrugBank compounds """
        self.save()
428
        # fingerprints to compute drugbank similarities are in settings module, default FP2
429
        fingerprinter = FingerPrinter(getattr(settings, "DRUGBANK_FINGERPRINTS", "FP2"))
430
431
432
433
434
435
436
437
438
        #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
439
440
441
    @property
    def biblio_refs(self):
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
442
        return all RefCompoundBiblio related to this compound
Hervé  MENAGER's avatar
Hervé MENAGER committed
443
444
        """
        return RefCompoundBiblio.objects.filter(compound=self)
445

446
447
448
449
450
451
452
453
454
455
456
457
458
    @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

459
    @property
Hervé  MENAGER's avatar
Hervé MENAGER committed
460
    def compound_action_ligand_ids(self):
461
462
463
        """
        return all PDB codes of the corresponding compound actions
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
464
        ligand_ids = set()
465
        for ca in self.compoundaction_set.all():
Hervé  MENAGER's avatar
Hervé MENAGER committed
466
467
            ligand_ids.add(ca.ligand_id)
        return ligand_ids
468

469
470
    @property
    def best_pXC50_activity(self):
471
        return self.compoundactivityresult_set.aggregate(Max('activity'))['activity__max']
472
473
474
475
476
477

    @property
    def best_pXC50_compound_activity_result(self):
        best_pXC50_activity = self.best_pXC50_activity
        if best_pXC50_activity is None:
            return None
478
        return self.compoundactivityresult_set.filter(activity=best_pXC50_activity)[0]
479

480
    @property
481
    def best_pXC50_activity_ppi_name(self):
482
        """
483
        Name of the PPI corresponding to the best PXC50 activity
484
        """
485
486
487
        best_activity_car = self.best_pXC50_compound_activity_result
        if best_activity_car is None:
            return None
488
489
490
        ppi_name = best_activity_car.test_activity_description.ppi.name
        return ppi_name

491
492
493
494
495
496
497
498
499
500
501
    @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

502
503
504
505
506
507
508
509
510
511
512
513
514
515
    @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()

516
517
518
519
520
521
    @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()]))
522
523
524
525
526
    
    @property
    def sorted_similar_drugbank_compounds(self):
        return self.drugbankcompoundtanimoto_set.order_by('-tanimoto')
        
527
528
529
    def autofill(self):
        # compute InChi and InChiKey
        self.inchi = smi2inchi(self.canonical_smile)
530
        self.inchikey = smi2inchikey(self.canonical_smile)
531
        self.compute_drugbank_compound_similarity()
532

533
534
535
    def __str__(self):
        return 'Compound #{}'.format(self.id)

536
537
538
class CompoundTanimoto(models.Model):
    canonical_smiles = models.TextField(
        'Canonical Smile')
539
    fingerprint = models.TextField('Fingerprint')
540
541
542
543
544
545
    compound = models.ForeignKey(Compound, models.CASCADE)
    tanimoto = models.DecimalField(
        'Tanimoto value', max_digits=5, decimal_places=4)

    class Meta:
        unique_together = (
546
            ('canonical_smiles', 'fingerprint', 'compound'))
547

548
def create_tanimoto(smiles_query, fingerprint):
549
    if CompoundTanimoto.objects.filter(canonical_smiles=smiles_query, fingerprint=fingerprint).count()==0:
550
        smiles_dict = {c.id:c.canonical_smile for c in Compound.objects.all()}
551
        fingerprinter = FingerPrinter(fingerprint)
552
553
554
555
556
        #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():
557
            cts.append(CompoundTanimoto(canonical_smiles=smiles_query, fingerprint=fingerprint, compound=Compound.objects.get(id=id_), tanimoto=tanimoto_dict[id_]))
558
        CompoundTanimoto.objects.bulk_create(cts)
559

560
class PcaBiplotData(models.Model):
561
562
    pca_biplot_data = models.TextField(
        'PCA biplot JSON data', blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
563

564

565
class LeLleBiplotData(models.Model):
566
567
    le_lle_biplot_data = models.TextField(
        'LE-LLE biplot JSON data', blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
568

569

570
571
class CellLine(models.Model):
    name = models.CharField('Name', max_length=50, unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
572

573
574
575
    def __str__(self):
        return self.name

Hervé  MENAGER's avatar
Hervé MENAGER committed
576

577
578
579
580
581
582
583
584
585
586
class TestActivityDescription(models.Model):
    TEST_TYPES = (
        ('BIOCH', 'Biochemical assay'),
        ('CELL', 'Cellular assay')
    )
    TEST_MODULATION_TYPES = (
        ('B', 'Binding'),
        ('I', 'Inhibition'),
        ('S', 'Stabilization')
    )
587
588
589
590
    PROTEIN_BOUND_CONSTRUCTS = (
        ('F', 'Full length'),
        ('U', 'Unspecified')
    )
591
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
592
593
    protein_domain_bound_complex = models.ForeignKey(
        ProteinDomainBoundComplex, models.CASCADE)
594
    ppi = models.ForeignKey(Ppi, models.CASCADE, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
595
    test_name = models.CharField('Test name', max_length=100)
596
    is_primary = models.BooleanField('Is primary')
Hervé  MENAGER's avatar
Hervé MENAGER committed
597
598
    protein_bound_construct = models.CharField(
        'Protein bound construct', max_length=5, choices=PROTEIN_BOUND_CONSTRUCTS, blank=True, null=True)
599
    test_type = models.CharField('Test type', max_length=5, choices=TEST_TYPES)
Hervé  MENAGER's avatar
Hervé MENAGER committed
600
601
602
603
    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
604
605
    cell_line = models.ForeignKey(
        CellLine, models.CASCADE, blank=True, null=True)
606

Hervé  MENAGER's avatar
Hervé MENAGER committed
607
    def get_complexes(self):
608
609
610
611
        """
        get the complexes tested for this PPI
        depends on the modulation type
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
612
        if self.test_modulation_type == 'I':
613
            return self.ppi.ppicomplex_set.all()
614
615
        else:
            return self.ppi.get_ppi_bound_complexes()
616

617
618
619
    @property
    def protein_domain_partner_complex(self):
        for ppic in self.ppi.ppicomplex_set.all():
Hervé  MENAGER's avatar
Hervé MENAGER committed
620
            if hasattr(ppic.complex, 'proteindomainpartnercomplex'):
621
622
623
                return ppic.complex.proteindomainpartnercomplex
        return None

624
625
626
627
628
629
630
    @property
    def name(self):
        return self.test_name

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

Hervé  MENAGER's avatar
Hervé MENAGER committed
631

632
class CompoundActivityResult(models.Model):
633
634
635
636
    MODULATION_TYPES = (
        ('I', 'Inhibition'),
        ('S', 'Stabilization')
    )
Hervé  MENAGER's avatar
Hervé MENAGER committed
637
    ACTIVITY_TYPES = (
Hervé  MENAGER's avatar
Hervé MENAGER committed
638
639
640
641
        ('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
642
    )
643
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
644
645
    test_activity_description = models.ForeignKey(
        TestActivityDescription, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
646
647
648
649
    activity_type = models.CharField(
        'Activity type', max_length=5, choices=ACTIVITY_TYPES)
    activity = models.DecimalField(
        'Activity', max_digits=12, decimal_places=10)
650
    inhibition_percentage = models.DecimalField(
651
        'Inhibition percentage', max_digits=3, decimal_places=0, null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
652
653
    modulation_type = models.CharField(
        'Modulation type', max_length=1, choices=MODULATION_TYPES)
Hervé  MENAGER's avatar
Hervé MENAGER committed
654
655

    class Meta:
Hervé  MENAGER's avatar
Hervé MENAGER committed
656
657
658
        unique_together = (
            ('compound', 'test_activity_description', 'activity_type'),)

659
660
    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
661

662
663
664
    def is_best(self):
        return self.compound.best_pXC50_compound_activity_result.id == self.id

Hervé  MENAGER's avatar
Hervé MENAGER committed
665

666
class TestCytotoxDescription(models.Model):
667
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
668
    test_name = models.CharField('Cytotoxicity test name', max_length=100)
669
    cell_line = models.ForeignKey(CellLine, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
670
671
672
    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
673

674
class CompoundCytotoxicityResult(models.Model):
675
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
676
677
    test_cytotoxicity_description = models.ForeignKey(
        TestCytotoxDescription, models.CASCADE)
678
    toxicity = models.BooleanField('Toxicity', default=False)
Hervé  MENAGER's avatar
Hervé MENAGER committed
679
680

    class Meta:
681
        unique_together = (('compound', 'test_cytotoxicity_description'),)
682

683
684
685
    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
686

687
class TestPKDescription(models.Model):
688
689
690
691
692
693
    ADMINISTRATION_MODES = (
        ('IV', ''),
        ('PO', ''),
        ('IP', ''),
        ('SL', 'SL')
    )
694
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
695
    test_name = models.CharField('Pharmacokinetic test name', max_length=100)
696
    organism = models.ForeignKey(Taxonomy, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
697
698
    administration_mode = models.CharField(
        'Administration mode', max_length=2, choices=ADMINISTRATION_MODES, blank=True, null=True)
699
700
    concentration = models.DecimalField(
        'Concentration in mg/l', max_digits=7, decimal_places=3, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
701
    dose = models.DecimalField(
702
        'Dose in mg/kg', max_digits=9, decimal_places=4, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
703
704
705
    dose_interval = models.IntegerField(
        'Dose interval, in hours', blank=True, null=True)

706
707

class CompoundPKResult(models.Model):
708
709
    compound = models.ForeignKey(Compound, models.CASCADE)
    test_pk_description = models.ForeignKey(TestPKDescription, models.CASCADE)
710
    tolerated = models.NullBooleanField('Tolerated', null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
711
712
713
714
    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)
715
    c_max = models.DecimalField(
Hervé  MENAGER's avatar
Hervé MENAGER committed
716
717
718
719
720
721
722
        '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
723
724

    class Meta:
725
        unique_together = (('compound', 'test_pk_description'),)
Hervé  MENAGER's avatar
Hervé MENAGER committed
726

727
728
729
    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
730

731
class CompoundAction(models.Model):
732
733
    ACTIVATION_MODES = (
        ('O', 'Orthosteric'),
734
735
        ('A', 'Allosteric'),
        ('U', 'Unspecified')
736
    )
737
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
738
739
    activation_mode = models.CharField(
        'Activation mode', max_length=1, choices=ACTIVATION_MODES)
740
    ppi = models.ForeignKey(Ppi, models.CASCADE)
741
    ligand_id = models.CharField('PDB Ligand ID', max_length=3, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
742
743
    nb_copy_compounds = models.IntegerField(
        'Number of copies for the compound')
Hervé  MENAGER's avatar
Hervé MENAGER committed
744
745

    class Meta:
746
        unique_together = (('ppi', 'compound', 'activation_mode', 'ligand_id'),)
Hervé  MENAGER's avatar
Hervé MENAGER committed
747
748

    def get_complexes(self):
749
750
751
752
        """
        get the complexes involved in the compound action
        which are always the bound complexes
        """
753
        return self.ppi.get_ppi_bound_complexes()
Hervé  MENAGER's avatar
Hervé MENAGER committed
754

755
756
    def __str__(self):
        return 'Action of {} on {}'.format(self.compound, self.ppi)
Hervé  MENAGER's avatar
Hervé MENAGER committed
757

758
class RefCompoundBiblio(models.Model):
759
760
    compound = models.ForeignKey(Compound, models.CASCADE)
    bibliography = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
761
762
    compound_name = models.CharField(
        'Compound name in the publication', max_length=50)
Hervé  MENAGER's avatar
Hervé MENAGER committed
763
764

    class Meta:
765
        unique_together = (('compound', 'bibliography'),)
766

767
class DrugBankCompound(models.Model):
768
769
770
771
772
    id = models.TextField(
        'Drugbank ID', unique=True, primary_key=True)
    common_name = models.TextField('Common name')
    canonical_smiles = models.TextField(
        'Canonical SMILES')
773
774
775
776
777

class DrugbankCompoundTanimoto(models.Model):
    compound = models.ForeignKey(Compound, models.CASCADE)
    drugbank_compound = models.ForeignKey(DrugBankCompound, models.CASCADE)
    tanimoto = models.DecimalField(
778
        'Tanimoto value', max_digits=5, decimal_places=4)