np.savetxt("./Data/smalldata/ML_B.csv", B, '%5.2f', delimiter=",") import lmm_lasso # hyperparameters for lasso mu = 1 min = -0.5 max = 0.5 numintv = 1000 rho = 1 alpha = 1.5 res = lmm_lasso.train(X, K, Y, mu=mu, numintervals=numintv, ldeltamin=min, ldeltamax=max, rho=rho, alpha=alpha) beta = res["weights"] print len(beta) np.savetxt("./Data/smalldata/lasso_B.csv", beta, '%5.2f', delimiter=",") np.savez("./Data/smalldata/ML", B_reml, B, beta) beta_true = beta_true.reshape((-1, )) beta_lasso = beta.reshape((-1, )) beta_ml = np.asarray(B).reshape((-1, )) beta_reml = np.asarray(B_reml).reshape((-1, )) choose = len(np.nonzero(beta_true)[0])
import sys # calculate the kinship (why?) K = lmm.calculateKinship(Z) # return beta, sigma # ML solution begin = time.time() B_reml = lmm.GWAS(Y, X, K) end = time.time() sys.stderr.write("Total time for 100 SNPs: %0.3f\n" % (end- begin)) # print B_reml np.savetxt(datapath + "REML_B.csv", B_reml, '%5.2f',delimiter=",") B_ml = lmm.GWAS(Y, X, K, REML=False) np.savetxt(datapath + "ML_B.csv", B_ml, '%5.2f',delimiter=",") import lmm_lasso res = lmm_lasso.train(X, K, Y, 0.5) beta = res["weights"] print len(beta) np.savetxt(datapath + "lasso_B.csv", beta, '%5.2f',delimiter=",") np.savez(datapath + "ML",B_reml,B_ml,beta) ################################################################# # draw ROC # ################################################################# beta_true = B.reshape((-1,)) beta_reml = np.asarray(B_reml) beta_ml = np.asarray(B_ml) beta_lasso = np.asarray(beta) beta_ml = beta_ml.reshape((-1,)) beta_lasso = beta_lasso.reshape((-1,)) beta_reml = beta_reml.reshape((-1,))
import sys # calculate the kinship (why?) K = lmm.calculateKinship(Z) # return beta, sigma # ML solution begin = time.time() B_reml = lmm.GWAS(Y, X, K) end = time.time() sys.stderr.write("Total time for 100 SNPs: %0.3f\n" % (end - begin)) # print B_reml np.savetxt(datapath + "REML_B.csv", B_reml, '%5.2f', delimiter=",") B_ml = lmm.GWAS(Y, X, K, REML=False) np.savetxt(datapath + "ML_B.csv", B_ml, '%5.2f', delimiter=",") import lmm_lasso res = lmm_lasso.train(X, K, Y, 0.5) beta = res["weights"] print len(beta) np.savetxt(datapath + "lasso_B.csv", beta, '%5.2f', delimiter=",") np.savez(datapath + "ML", B_reml, B_ml, beta) ################################################################# # draw ROC # ################################################################# beta_true = B.reshape((-1, )) beta_reml = np.asarray(B_reml) beta_ml = np.asarray(B_ml) beta_lasso = np.asarray(beta) beta_ml = beta_ml.reshape((-1, )) beta_lasso = beta_lasso.reshape((-1, )) beta_reml = beta_reml.reshape((-1, ))
sys.stderr.write("Total time for 100 SNPs: %0.3f\n" % (end- begin)) # print B_reml np.savetxt("./Data/smalldata/REML_B.csv", B_reml, '%5.2f',delimiter=",") B = lmm.GWAS(Y, X, K, REML=False) np.savetxt("./Data/smalldata/ML_B.csv", B, '%5.2f',delimiter=",") import lmm_lasso # hyperparameters for lasso mu = 1 min = -0.5 max = 0.5 numintv = 1000 rho= 1 alpha= 1.5 res = lmm_lasso.train(X, K, Y, mu=mu, numintervals=numintv, ldeltamin=min, ldeltamax=max, rho=rho, alpha=alpha) beta = res["weights"] print len(beta) np.savetxt("./Data/smalldata/lasso_B.csv", beta, '%5.2f',delimiter=",") np.savez("./Data/smalldata/ML",B_reml,B,beta) beta_true = beta_true.reshape((-1,)) beta_lasso = beta.reshape((-1,)) beta_ml = np.asarray(B).reshape((-1,)) beta_reml = np.asarray(B_reml).reshape((-1,)) choose = len(np.nonzero(beta_true)[0]) beta_ml[np.argsort(beta_ml)[:(np.shape(beta_true)[0]-choose)]] = 0 beta_reml[np.argsort(beta_reml)[:(np.shape(beta_true)[0]-choose)]] = 0