models.py 30.8 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
    def __str__(self):
        return self.description

Hervé  MENAGER's avatar
Hervé MENAGER committed
110

111
class Protein(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
112
    uniprot_id = models.CharField('Uniprot ID', unique=True, max_length=10)
Hervé  MENAGER's avatar
Hervé MENAGER committed
113
114
    recommended_name_long = models.CharField(
        'Uniprot Recommended Name (long)', max_length=75)
Hervé  MENAGER's avatar
Hervé MENAGER committed
115
116
117
    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)
118
    organism = models.ForeignKey('Taxonomy', models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
119
120
    molecular_functions = models.ManyToManyField(MolecularFunction)

121
    def autofill(self):
122
        info = get_uniprot_info(self.uniprot_id)
123
        self.recommended_name_long = info['recommended_name']
124
125
        self.gene_name = info['gene']
        self.entry_name = info['entry_name']
126
        self.short_name = info['short_name']
127
128
129
130
131
        try:
            taxonomy = Taxonomy.objects.get(taxonomy_id=info['organism'])
        except Taxonomy.DoesNotExist:
            taxonomy = Taxonomy()
            taxonomy.taxonomy_id = info['organism']
132
            taxonomy.save(autofill=True)
133
        self.organism = taxonomy
134
        super(Protein, self).save()
135
136
137
138
139
140
        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
141
                mol_function.save(autofill=True)
142
            self.molecular_functions.add(mol_function)
143

144
145
146
    def __str__(self):
        return '{} ({})'.format(self.uniprot_id, self.recommended_name_long)

Hervé  MENAGER's avatar
Hervé MENAGER committed
147

148
class Domain(AutoFillableModel):
Hervé  MENAGER's avatar
Hervé MENAGER committed
149
150
    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
151
    pfam_description = models.CharField('Pfam Description', max_length=100)
Hervé  MENAGER's avatar
Hervé MENAGER committed
152
    domain_family = models.CharField('Domain family', max_length=25)
Hervé  MENAGER's avatar
Hervé MENAGER committed
153
154
    # TODO: what is this field? check database
    # contents
155

156
    def autofill(self):
157
158
159
        info = get_pfam_info(self.pfam_acc)
        self.pfam_id = info['id']
        self.pfam_description = info['description']
Hervé  MENAGER's avatar
Hervé MENAGER committed
160

161
162
163
    def __str__(self):
        return '{} ({}-{})'.format(self.pfam_acc, self.pfam_id, self.pfam_description)

Hervé  MENAGER's avatar
Hervé MENAGER committed
164

165
class ProteinDomainComplex(models.Model):
166
167
    protein = models.ForeignKey('Protein', models.CASCADE)
    domain = models.ForeignKey('Domain', models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
168
169
170
    ppc_copy_nb = models.IntegerField(
        'Number of copies of the protein in the complex')

Hervé  MENAGER's avatar
Hervé MENAGER committed
171
172
    class Meta:
        verbose_name_plural = "complexes"
173

174
175
176
    def __str__(self):
        return '{}-{}'.format(self.protein_id, self.domain_id)

177
178
    def name(self):
        return self.protein.short_name
Hervé  MENAGER's avatar
Hervé MENAGER committed
179

180

181
class ProteinDomainBoundComplex(ProteinDomainComplex):
Hervé  MENAGER's avatar
Hervé MENAGER committed
182
183
184
    ppp_copy_nb_per_p = models.IntegerField(
        'Number of copies of the protein in the pocket')

185
186
    class Meta:
        verbose_name_plural = "bound complexes"
Hervé  MENAGER's avatar
Hervé MENAGER committed
187
188


189
class ProteinDomainPartnerComplex(ProteinDomainComplex):
Hervé  MENAGER's avatar
Hervé MENAGER committed
190

191
192
    class Meta:
        verbose_name_plural = "partner complexes"
Hervé  MENAGER's avatar
Hervé MENAGER committed
193

Hervé  MENAGER's avatar
Hervé MENAGER committed
194

195
196
197
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
198

199
200
201
    class Meta:
        verbose_name_plural = "symmetries"

202
203
204
    def __str__(self):
        return '{} ({})'.format(self.code, self.description)

205
206

class Disease(models.Model):
Hervé  MENAGER's avatar
Hervé MENAGER committed
207
    name = models.CharField('Disease', max_length=30, unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
208
    # is there any database/nomenclature for diseases?
209
210
211
212

    def __str__(self):
        return self.name

Hervé  MENAGER's avatar
Hervé MENAGER committed
213

Hervé  MENAGER's avatar
Hervé MENAGER committed
214
215
216
class PpiFamily(models.Model):
    name = models.CharField('Name', max_length=30, unique=True)

217
218
219
    class Meta:
        verbose_name_plural = "PPI Families"

Hervé  MENAGER's avatar
Hervé MENAGER committed
220
221
    def __str__(self):
        return self.name
Hervé  MENAGER's avatar
Hervé MENAGER committed
222

Hervé  MENAGER's avatar
Hervé MENAGER committed
223

224
class Ppi(AutoFillableModel):
225
    pdb_id = models.CharField('PDB ID', max_length=4, null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
226
227
    pockets_nb = models.IntegerField(
        'Total number of pockets in the complex', default=1)
228
    symmetry = models.ForeignKey(Symmetry, models.CASCADE)
229
    diseases = models.ManyToManyField(Disease)
230
    family = models.ForeignKey(PpiFamily, models.CASCADE, null=True, blank=True)
231
    name = models.TextField('PPI name', null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
232

233
    def __str__(self):
234
        return 'PPI #{} on {}'.format(self.id, self.name)
235

236
237
238
    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()
239
240
241
242
243

    def get_ppi_bound_complexes(self):
        """
        return bound ppi complexes belonging to this ppi
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
244
        # this is the less efficient query ever seen, FIXME
245
246
        return PpiComplex.objects.filter(ppi=self, complex__in=ProteinDomainBoundComplex.objects.all())

247
    def compute_name_from_protein_names(self):
Hervé  MENAGER's avatar
Hervé MENAGER committed
248
        all_protein_names = set(
249
            [ppi_complex.complex.protein.short_name for ppi_complex in self.ppicomplex_set.all()])
Hervé  MENAGER's avatar
Hervé MENAGER committed
250
251
        bound_protein_names = set(
            [ppi_complex.complex.protein.short_name for ppi_complex in self.get_ppi_bound_complexes()])
252
253
254
255
        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
256
        if partner_str != '':
257
258
            name += ' / ' + partner_str
        return name
259

Hervé  MENAGER's avatar
Hervé MENAGER committed
260

Hervé  MENAGER's avatar
Hervé MENAGER committed
261
class PpiComplex(models.Model):
262
263
    ppi = models.ForeignKey(Ppi, models.CASCADE)
    complex = models.ForeignKey(ProteinDomainComplex, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
264
265
    cc_nb = models.IntegerField(
        'Number of copies of the complex in the PPI', default=1)
Hervé  MENAGER's avatar
Hervé MENAGER committed
266
267
268
269

    class Meta:
        verbose_name_plural = "Ppi complexes"

270
271
272
    def __str__(self):
        return 'PPI {}, Complex {} ({})'.format(self.ppi, self.complex, self.cc_nb)

Hervé  MENAGER's avatar
Hervé MENAGER committed
273

274
275
276
277
class CompoundManager(models.Manager):

    def get_queryset(self):
        qs = super().get_queryset()
278
        # with number of publications
279
        qs = qs.annotate(pubs=Count('refcompoundbiblio', distinct=True))
280
        # with best activity
281
        qs = qs.annotate(best_activity=Max('compoundactivityresult__activity')) 
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
        # 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))
306
307
        #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()))
308
309
310
311
312
        # 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)
313
314
        #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()))
315
316
        # PDB ligand available
        qs = qs.annotate(pdb_ligand_av=Case(When(compoundaction__ppi__pdb_id__isnull=False, then=True), default=False, output_field=BooleanField()))
317
318
319
320
321
322
        # 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()))
323
324
        return qs

325
class Compound(AutoFillableModel):
326
    objects = CompoundManager() 
327
328
    canonical_smile = models.TextField(
        'Canonical Smile', unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
329
330
331
332
333
    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)
334
    fsp3 = models.DecimalField('Fsp3', max_digits=3, decimal_places=2)
Hervé  MENAGER's avatar
Hervé MENAGER committed
335
336
337
338
339
340
341
342
343
344
    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)
345
346
347
348
    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
349
350
351
352
    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')
353
354
    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
355
356
357
358
359
360
361
362
363
364
365
366
367
368
    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')
369
    nb_n = models.IntegerField('Number of nitrogen atoms')
Hervé  MENAGER's avatar
Hervé MENAGER committed
370
371
372
    nb_o = models.IntegerField('Number of oxygen atoms')
    nb_rings = models.IntegerField('Number of rings')
    nb_rotatable_bonds = models.IntegerField('Number of rotatable bonds')
373
374
    inchi = models.TextField('InChi')
    inchikey = models.TextField('InChiKey')
Hervé  MENAGER's avatar
Hervé MENAGER committed
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
    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)
400

401
402
403
    class Meta:
       ordering = ['id']

404
405
406
    def compute_drugbank_compound_similarity(self):
        """ compute Tanimoto similarity to existing DrugBank compounds """
        self.save()
407
        # fingerprints to compute drugbank similarities are in settings module, default FP2
408
        fingerprinter = FingerPrinter(getattr(settings, "DRUGBANK_FINGERPRINTS", "FP2"))
409
410
411
412
413
414
415
416
417
        #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
418
419
420
    @property
    def biblio_refs(self):
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
421
        return all RefCompoundBiblio related to this compound
Hervé  MENAGER's avatar
Hervé MENAGER committed
422
423
        """
        return RefCompoundBiblio.objects.filter(compound=self)
424

425
426
427
428
429
430
431
432
433
434
435
436
437
    @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

438
    @property
Hervé  MENAGER's avatar
Hervé MENAGER committed
439
    def compound_action_ligand_ids(self):
440
441
442
        """
        return all PDB codes of the corresponding compound actions
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
443
        ligand_ids = set()
444
        for ca in self.compoundaction_set.all():
Hervé  MENAGER's avatar
Hervé MENAGER committed
445
446
            ligand_ids.add(ca.ligand_id)
        return ligand_ids
447

448
449
    @property
    def best_pXC50_activity(self):
450
        return self.compoundactivityresult_set.aggregate(Max('activity'))['activity__max']
451
452
453
454
455
456

    @property
    def best_pXC50_compound_activity_result(self):
        best_pXC50_activity = self.best_pXC50_activity
        if best_pXC50_activity is None:
            return None
457
        return self.compoundactivityresult_set.filter(activity=best_pXC50_activity)[0]
458

459
    @property
460
    def best_pXC50_activity_ppi_name(self):
461
        """
462
        Name of the PPI corresponding to the best PXC50 activity
463
        """
464
465
466
        best_activity_car = self.best_pXC50_compound_activity_result
        if best_activity_car is None:
            return None
467
468
469
        ppi_name = best_activity_car.test_activity_description.ppi.name
        return ppi_name

470
471
472
473
474
475
476
477
478
479
480
    @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

481
482
483
484
485
486
487
488
489
490
491
492
493
494
    @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()

495
496
497
498
499
500
    @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()]))
501
502
503
504
505
    
    @property
    def sorted_similar_drugbank_compounds(self):
        return self.drugbankcompoundtanimoto_set.order_by('-tanimoto')
        
506
507
508
    def autofill(self):
        # compute InChi and InChiKey
        self.inchi = smi2inchi(self.canonical_smile)
509
        self.inchikey = smi2inchikey(self.canonical_smile)
510
        self.compute_drugbank_compound_similarity()
511

512
513
514
    def __str__(self):
        return 'Compound #{}'.format(self.id)

515
516
517
class CompoundTanimoto(models.Model):
    canonical_smiles = models.TextField(
        'Canonical Smile')
518
    fingerprint = models.TextField('Fingerprint')
519
520
521
522
523
524
    compound = models.ForeignKey(Compound, models.CASCADE)
    tanimoto = models.DecimalField(
        'Tanimoto value', max_digits=5, decimal_places=4)

    class Meta:
        unique_together = (
525
            ('canonical_smiles', 'fingerprint', 'compound'))
526

527
def create_tanimoto(smiles_query, fingerprint):
528
    if CompoundTanimoto.objects.filter(canonical_smiles=smiles_query, fingerprint=fingerprint).count()==0:
529
        smiles_dict = {c.id:c.canonical_smile for c in Compound.objects.all()}
530
        fingerprinter = FingerPrinter(fingerprint)
531
532
533
534
535
        #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():
536
            cts.append(CompoundTanimoto(canonical_smiles=smiles_query, fingerprint=fingerprint, compound=Compound.objects.get(id=id_), tanimoto=tanimoto_dict[id_]))
537
        CompoundTanimoto.objects.bulk_create(cts)
538

539
class PcaBiplotData(models.Model):
540
541
    pca_biplot_data = models.TextField(
        'PCA biplot JSON data', blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
542

543

544
class LeLleBiplotData(models.Model):
545
546
    le_lle_biplot_data = models.TextField(
        'LE-LLE biplot JSON data', blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
547

548

549
550
class CellLine(models.Model):
    name = models.CharField('Name', max_length=50, unique=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
551

552
553
554
    def __str__(self):
        return self.name

Hervé  MENAGER's avatar
Hervé MENAGER committed
555

556
557
558
559
560
561
562
563
564
565
class TestActivityDescription(models.Model):
    TEST_TYPES = (
        ('BIOCH', 'Biochemical assay'),
        ('CELL', 'Cellular assay')
    )
    TEST_MODULATION_TYPES = (
        ('B', 'Binding'),
        ('I', 'Inhibition'),
        ('S', 'Stabilization')
    )
566
567
568
569
    PROTEIN_BOUND_CONSTRUCTS = (
        ('F', 'Full length'),
        ('U', 'Unspecified')
    )
570
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
571
572
    protein_domain_bound_complex = models.ForeignKey(
        ProteinDomainBoundComplex, models.CASCADE)
573
    ppi = models.ForeignKey(Ppi, models.CASCADE, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
574
    test_name = models.CharField('Test name', max_length=100)
575
    is_primary = models.BooleanField('Is primary')
Hervé  MENAGER's avatar
Hervé MENAGER committed
576
577
    protein_bound_construct = models.CharField(
        'Protein bound construct', max_length=5, choices=PROTEIN_BOUND_CONSTRUCTS, blank=True, null=True)
578
    test_type = models.CharField('Test type', max_length=5, choices=TEST_TYPES)
Hervé  MENAGER's avatar
Hervé MENAGER committed
579
580
581
582
    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
583
584
    cell_line = models.ForeignKey(
        CellLine, models.CASCADE, blank=True, null=True)
585

Hervé  MENAGER's avatar
Hervé MENAGER committed
586
    def get_complexes(self):
587
588
589
590
        """
        get the complexes tested for this PPI
        depends on the modulation type
        """
Hervé  MENAGER's avatar
Hervé MENAGER committed
591
        if self.test_modulation_type == 'I':
592
            return self.ppi.ppicomplex_set.all()
593
594
        else:
            return self.ppi.get_ppi_bound_complexes()
595

596
597
598
    @property
    def protein_domain_partner_complex(self):
        for ppic in self.ppi.ppicomplex_set.all():
Hervé  MENAGER's avatar
Hervé MENAGER committed
599
            if hasattr(ppic.complex, 'proteindomainpartnercomplex'):
600
601
602
                return ppic.complex.proteindomainpartnercomplex
        return None

Hervé  MENAGER's avatar
Hervé MENAGER committed
603

604
class CompoundActivityResult(models.Model):
605
606
607
608
609
    MODULATION_TYPES = (
        ('B', 'Binding'),
        ('I', 'Inhibition'),
        ('S', 'Stabilization')
    )
Hervé  MENAGER's avatar
Hervé MENAGER committed
610
    ACTIVITY_TYPES = (
Hervé  MENAGER's avatar
Hervé MENAGER committed
611
612
613
614
        ('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
615
    )
616
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
617
618
    test_activity_description = models.ForeignKey(
        TestActivityDescription, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
619
620
621
622
    activity_type = models.CharField(
        'Activity type', max_length=5, choices=ACTIVITY_TYPES)
    activity = models.DecimalField(
        'Activity', max_digits=12, decimal_places=10)
623
    inhibition_percentage = models.DecimalField(
624
        'Inhibition percentage', max_digits=3, decimal_places=0, null=True, blank=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
625
626
    modulation_type = models.CharField(
        'Modulation type', max_length=1, choices=MODULATION_TYPES)
Hervé  MENAGER's avatar
Hervé MENAGER committed
627
628

    class Meta:
Hervé  MENAGER's avatar
Hervé MENAGER committed
629
630
631
        unique_together = (
            ('compound', 'test_activity_description', 'activity_type'),)

632
633
    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
634

635
636
637
    def is_best(self):
        return self.compound.best_pXC50_compound_activity_result.id == self.id

Hervé  MENAGER's avatar
Hervé MENAGER committed
638

639
class TestCytotoxDescription(models.Model):
640
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
641
    test_name = models.CharField('Cytotoxicity test name', max_length=100)
642
    cell_line = models.ForeignKey(CellLine, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
643
644
645
    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
646

647
class CompoundCytotoxicityResult(models.Model):
648
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
649
650
    test_cytotoxicity_description = models.ForeignKey(
        TestCytotoxDescription, models.CASCADE)
651
    toxicity = models.BooleanField('Toxicity', default=False)
Hervé  MENAGER's avatar
Hervé MENAGER committed
652
653

    class Meta:
654
        unique_together = (('compound', 'test_cytotoxicity_description'),)
655

656
657
658
    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
659

660
class TestPKDescription(models.Model):
661
662
663
664
665
666
    ADMINISTRATION_MODES = (
        ('IV', ''),
        ('PO', ''),
        ('IP', ''),
        ('SL', 'SL')
    )
667
    biblio = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
668
    test_name = models.CharField('Pharmacokinetic test name', max_length=100)
669
    organism = models.ForeignKey(Taxonomy, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
670
671
    administration_mode = models.CharField(
        'Administration mode', max_length=2, choices=ADMINISTRATION_MODES, blank=True, null=True)
672
673
    concentration = models.DecimalField(
        'Concentration in mg/l', max_digits=7, decimal_places=3, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
674
    dose = models.DecimalField(
675
        'Dose in mg/kg', max_digits=9, decimal_places=4, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
676
677
678
    dose_interval = models.IntegerField(
        'Dose interval, in hours', blank=True, null=True)

679
680

class CompoundPKResult(models.Model):
681
682
    compound = models.ForeignKey(Compound, models.CASCADE)
    test_pk_description = models.ForeignKey(TestPKDescription, models.CASCADE)
683
    tolerated = models.NullBooleanField('Tolerated', null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
684
685
686
687
    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)
688
    c_max = models.DecimalField(
Hervé  MENAGER's avatar
Hervé MENAGER committed
689
690
691
692
693
694
695
        '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
696
697

    class Meta:
698
        unique_together = (('compound', 'test_pk_description'),)
Hervé  MENAGER's avatar
Hervé MENAGER committed
699

700
701
702
    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
703

704
class CompoundAction(models.Model):
705
706
    ACTIVATION_MODES = (
        ('O', 'Orthosteric'),
707
708
        ('A', 'Allosteric'),
        ('U', 'Unspecified')
709
    )
710
    compound = models.ForeignKey(Compound, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
711
712
    activation_mode = models.CharField(
        'Activation mode', max_length=1, choices=ACTIVATION_MODES)
713
    ppi = models.ForeignKey(Ppi, models.CASCADE)
714
    ligand_id = models.CharField('PDB Ligand ID', max_length=3, blank=True, null=True)
Hervé  MENAGER's avatar
Hervé MENAGER committed
715
716
    nb_copy_compounds = models.IntegerField(
        'Number of copies for the compound')
Hervé  MENAGER's avatar
Hervé MENAGER committed
717
718

    class Meta:
719
        unique_together = (('ppi', 'compound', 'activation_mode', 'ligand_id'),)
Hervé  MENAGER's avatar
Hervé MENAGER committed
720
721

    def get_complexes(self):
722
723
724
725
        """
        get the complexes involved in the compound action
        which are always the bound complexes
        """
726
        return self.ppi.get_ppi_bound_complexes()
Hervé  MENAGER's avatar
Hervé MENAGER committed
727

728
729
    def __str__(self):
        return 'Action of {} on {}'.format(self.compound, self.ppi)
Hervé  MENAGER's avatar
Hervé MENAGER committed
730

731
class RefCompoundBiblio(models.Model):
732
733
    compound = models.ForeignKey(Compound, models.CASCADE)
    bibliography = models.ForeignKey(Bibliography, models.CASCADE)
Hervé  MENAGER's avatar
Hervé MENAGER committed
734
735
    compound_name = models.CharField(
        'Compound name in the publication', max_length=50)
Hervé  MENAGER's avatar
Hervé MENAGER committed
736
737

    class Meta:
738
        unique_together = (('compound', 'bibliography'),)
739

740
class DrugBankCompound(models.Model):
741
742
743
744
745
    id = models.TextField(
        'Drugbank ID', unique=True, primary_key=True)
    common_name = models.TextField('Common name')
    canonical_smiles = models.TextField(
        'Canonical SMILES')
746
747
748
749
750

class DrugbankCompoundTanimoto(models.Model):
    compound = models.ForeignKey(Compound, models.CASCADE)
    drugbank_compound = models.ForeignKey(DrugBankCompound, models.CASCADE)
    tanimoto = models.DecimalField(
751
        'Tanimoto value', max_digits=5, decimal_places=4)