def evalCorrAMI(root):
# Axx.json
# Bxx.json
# AX_BX_clip
# AX_BX_rec_a
# AX_BX_rec_b
namelist = utils.getNames(root)
a_array = utils.json2numpy(root + "/" + namelist[0])
a_rec = utils.json2numpy(root + "/" + namelist[3])
b_array = utils.json2numpy(root + "/" + namelist[1])
b_rec = utils.json2numpy(root + "/" + namelist[4])
Corr_a_ap, Corr_b_ap, Corr_a_bp, Corr_b_bp = \
Corr(a_array, a_rec), Corr(b_array, a_rec), Corr(a_array, b_rec), Corr(b_array, b_rec)
AMI = Corr_a_ap - Corr_b_ap - Corr_a_bp + Corr_b_bp
return Corr_a_ap, Corr_b_ap, Corr_a_bp, Corr_b_bp, AMI, Corr_a_ap - Corr_b_ap, Corr_b_bp - Corr_a_bp
#Open a memory mapped distance matrix.
#We do this because the pairwise distance matrix for 100 targets does not fit in memory.
#It is nearly 100% dense and has 117747*117747 = 13864356009 elements. This is also
#Why it uses float16 (reducing the required storage space to ~26GB, c.f. 52GB for float32).
distance_matrix = np.memmap('./processed_data/graph_fp_comparison/distMat.dat', dtype=np.float16,
shape=(x_.shape[0], x_.shape[0]))
#build the hierarchical clustering tree:
clusterer = ParisClusterer(x_.toarray())
clusterer.buildAdjacency()
clusterer.fit()
#Fingerprints to be compared:
fp_names = utils.getNames()
fp_dict = {} #this stores the actual fingerprint feature matrices
fp_ap_before_trim = {} #this stores the average precisions
fp_ap_after_trim = {} #this stores the average precisions
#Load up the dictionaries with the relevant feature matrices for each fingerprint:
for fp in fp_names:
print(fp)
featureMatrix, labels = utils.load_feature_and_label_matrices(type=fp)
featureMatrix_, labels__ = utils.get_subset(featureMatrix, y, indices=col_indices)
fp_dict[fp]=sparse.csr_matrix(featureMatrix_)
fp_ap_before_trim[fp] = []
fp_ap_after_trim[fp] = []
#Store all the results in these:
import utils from sklearn.metrics import precision_score, recall_score, roc_auc_score, label_ranking_average_precision_score from sklearn.metrics import label_ranking_loss, confusion_matrix, average_precision_score, auc, precision_recall_curve import statsmodels.api as sm from statsmodels.distributions.empirical_distribution import ECDF from tqdm import tqdm from seaborn import kdeplot import pymc3 as pm ##Set plotting parameters: utils.set_mpl_params() fp_names = utils.getNames(short=False) nicenames = dict() nicenames['morgan'] = 'Morgan' nicenames['2dpharm'] = '2D\nPharm.' nicenames['atom_pair'] = 'Atom Pair' nicenames['erg'] = 'ErG' nicenames['cats'] = 'CATS' nicenames['layered'] = 'Layered' nicenames['maccs'] = 'MACCS' nicenames['morgan_feat'] = 'Morgan\nFeat.' nicenames['pattern'] = 'Pattern' nicenames['rdk'] = 'RDK' nicenames['topo_torsion'] = 'Topol.\nTorsion' fp_aps_before = dict()
import utils
import random
NAMES = utils.getNames()
EMAILS = utils.getEmails()
CITY = utils.getCity()
STATE = utils.getState()
PHONE = utils.getPhone()
DATAS_USERS = []
def shuffleDatas():
random.shuffle(NAMES)
random.shuffle(EMAILS)
random.shuffle(CITY)
random.shuffle(STATE)
random.shuffle(PHONE)
def controller():
utils.showProgramName()
utils.line()
# ---------------
stop = False
while stop == False:
utils.menuOptions()
shuffleDatas()
optionUser = str(input('Selecione uma ou mais opções:'))
optionUser = optionUser.split(',')
for option in optionUser: