F = 1.*(SP.rand(N,2)<0.2); A = SP.eye(P)
# add second fixed effect
F2 = 1.*(SP.rand(N,3)<0.2); A2 = SP.ones((1,P))
mu.addFixedEffect(F=F,A=A)
mu.addFixedEffect(F=F2,A=A2)
if mtSet_present:
mu1 = MEAN.mean(Y)
mu1.addFixedEffect(F=F,A=A)
mu1.addFixedEffect(F=F2,A=A2)
# define covariance matrices
Cg = freeform(P)
Cn = freeform(P)
Cg1 = freeform(P)
Cn1 = freeform(P)
ipdb.set_trace()
if 1:
# compare with mtSet implementation
params = {}
params['Cg'] = SP.randn(int(0.5*P*(P+1)))
params['Cn'] = SP.randn(int(0.5*P*(P+1)))
print "check gradient with gp2kronSum"
gp = gp2kronSum(mu,Cg,Cn,XX)
# import data
fname = 'data/arab107_preprocessed.hdf5'
f = h5py.File(fname,'r')
X = f['genotype'][:]
X-= X.mean(0); X/=X.std(0)
G = X[:,0:200]
Y = genPheno(G=G,X=X,var_g=0.10,var_x=0.40,P=P)
XX = SP.dot(X,X.T)
XX/= XX.diagonal().mean()
XX+= 1e-4*SP.eye(XX.shape[0])
GG = SP.dot(G,G.T)
GG/= GG.diagonal().mean()
GG+= 1e-4*SP.eye(GG.shape[0])
C1 = freeform(P)
C2 = freeform(P)
Cn = freeform(P)
gp = gp3kronSumApprox(Y=Y,C1=C1,C2=C2,Cn=Cn,R1=GG,R2=XX,tol=1e-16)
if 0:
Kinit = SP.cov(Y.T)/3
C1.setCovariance(Kinit)
C2.setCovariance(Kinit)
Cn.setCovariance(Kinit)
else:
C1.setRandomParams()
C2.setRandomParams()
Cn.setRandomParams()
params = gp.getParams()
gp.setParams(params)
# define mean term
mu = mean(Y)
# add any fixed effect including bias term
F = 1. * (SP.rand(N, 3) < 0.2)
A = SP.eye(P)
F[:, 0] = 1.0
mu.addFixedEffect(F=F, A=A)
# add common fixed effect
F2 = 1. * (SP.rand(N, 4) < 0.2)
A2 = SP.ones((1, P))
mu.addFixedEffect(F=F2, A=A2)
# define covariance matrices
Cg = freeform(P)
Cn = freeform(P)
Cg1 = freeform(P)
Cn1 = freeform(P)
# compare with mtSet implementation
params = {}
params['Cg'] = SP.randn(int(0.5 * P * (P + 1)))
params['Cn'] = SP.randn(int(0.5 * P * (P + 1)))
print "creating gp2kronSum object"
XX = SP.dot(X, X.T)
XX /= XX.diagonal().mean()
gp = gp2kronSum(mu, Cg, Cn, XX)
gp.setParams(params)
fname = 'data/arab107_preprocessed.hdf5'
f = h5py.File(fname, 'r')
X = f['genotype'][:]
X -= X.mean(0)
X /= X.std(0)
G = X[:, 0:200]
Y = genPheno(G=G, X=X, var_g=0.10, var_x=0.40, P=P)
XX = SP.dot(X, X.T)
XX /= XX.diagonal().mean()
XX += 1e-4 * SP.eye(XX.shape[0])
GG = SP.dot(G, G.T)
GG /= GG.diagonal().mean()
GG += 1e-4 * SP.eye(GG.shape[0])
Cr = freeform(P)
Cg = freeform(P)
Cn = freeform(P)
gp = gp3kronSumApprox(Y=Y, Cr=Cr, Cg=Cg, Cn=Cn, XX=XX, GG=GG)
n_rips = 10
for rip in range(n_rips):
Cr.setRandomParams()
Cg.setRandomParams()
Cn.setRandomParams()
params = gp.getParams()
gp.setParams(params)
conv, info = OPT.opt_hyper(gp, params, factr=1e3)
print conv
# import data
fname = 'data/arab107_preprocessed.hdf5'
f = h5py.File(fname,'r')
X = f['genotype'][:]
X-= X.mean(0); X/=X.std(0)
G = X[:,0:200]
Y = genPheno(G=G,X=X,var_g=0.10,var_x=0.40,P=P)
XX = SP.dot(X,X.T)
XX/= XX.diagonal().mean()
XX+= 1e-4*SP.eye(XX.shape[0])
GG = SP.dot(G,G.T)
GG/= GG.diagonal().mean()
GG+= 1e-4*SP.eye(GG.shape[0])
Cr = freeform(P)
Cg = freeform(P)
Cn = freeform(P)
gp = gp3kronSumApprox(Y=Y,Cr=Cr,Cg=Cg,Cn=Cn,XX=XX,GG=GG)
n_rips = 10
for rip in range(n_rips):
Cr.setRandomParams()
Cg.setRandomParams()
Cn.setRandomParams()
params = gp.getParams()
gp.setParams(params)
conv,info = OPT.opt_hyper(gp,params,factr=1e3)
print conv
# import data
fname = 'data/arab107_preprocessed.hdf5'
f = h5py.File(fname,'r')
X = f['genotype'][:]
X-= X.mean(0); X/=X.std(0)
G = X[:,0:200]
Y = genPheno(G=G,X=X,var_g=0.10,var_x=0.40,P=P)
XX = SP.dot(X,X.T)
XX/= XX.diagonal().mean()
XX+= 1e-4*SP.eye(XX.shape[0])
GG = SP.dot(G,G.T)
GG/= GG.diagonal().mean()
GG+= 1e-4*SP.eye(GG.shape[0])
C1 = freeform(P)
C2 = freeform(P)
Cn = freeform(P)
gp = gp3kronSumApprox(Y=Y,C1=C1,C2=C2,Cn=Cn,R1=GG,R2=XX,tol=1e-16)
if 0:
Kinit = SP.cov(Y.T)/3
C1.setCovariance(Kinit)
C2.setCovariance(Kinit)
Cn.setCovariance(Kinit)
else:
C1.setRandomParams()
C2.setRandomParams()
Cn.setRandomParams()
params = gp.getParams()
gp.setParams(params)