def goodness_of_fit(ddGs): """ Performs Shapiro-Wilk test on array of ddGs Parameters: ddGs (array-like): Tabulation of all 19 ddGs for all residues of a given protein Returns: ks (list of float): Test statistic of SW test for all residues pvalues (list of float): Pvalue of SW test for all residues """ fit_gaussian_mixture = FitGaussianMixture(ddGs) ks = [] pvalues = [] for res, ddGs_res in enumerate(ddGs): ddGs_19 = fit_gaussian_mixture.preprocess(ddGs_res) k, pvalue = stats.shapiro(ddGs_19) ks.append(k) pvalues.append(pvalue) return ks, pvalues
def plotout(protein_name, rejected_contacts, all_contacts):
plt.title(protein_name)
plt.boxplot([rejected_contacts, all_contacts],
showmeans=True) #, label = 'all')
plt.xlabel('residues')
plt.xticks([1, 2], ['reject the null', 'all'])
plt.ylabel('frequency of residues')
k, pvalue = stats.ks_2samp(all_contacts,
rejected_contacts) #only CAII significant
plt.legend(title='KS = {0:.2f} ({1:.2E})'.format(k, pvalue))
plt.show()
if __name__ == "__main__":
ddGs_exp = read_in_experiment() #ordered by res-1
fit_gaussian_mixture = FitGaussianMixture(ddGs_exp)
ks = []
pvalues = []
residues_ref = []
for res, ddGs_res in enumerate(ddGs_exp):
ddGs_19 = fit_gaussian_mixture.preprocess(ddGs_res)
if len(ddGs_19) == 17:
k, pvalue = stats.shapiro(ddGs_19)
ks.append(k)
pvalues.append(pvalue)
residues_ref.append(res + 1) #account for res-1
residues_fail = np.where(np.array(pvalues) < 0.05)[0]
d_contacts = get_contacts('1pga')
linestyle='--',
color='k')
plt.xticks(x, pretty_proteins) # rotation = 20)
plt.ylabel('K-S test statistic')
plt.legend()
plt.show()
if __name__ == '__main__':
data_file = 'Tokuriki_2007.xlsx'
df = pd.read_excel(data_file)
ks = []
ks2 = []
for protein_name in proteins:
protein = df[df['Name'] == protein_name]
ddGs = get_protein_ddG(protein)
fit_gaussian_mixture = FitGaussianMixture(ddGs)
ddGs_overall = fit_gaussian_mixture.preprocess(ddGs) #destroys shape
yPred = generate_dist(ddGs, fit_gaussian_mixture)
k, pvalue = stats.ks_2samp(fit_gaussian_mixture.hist, yPred)
ks.append(k)
ks2.append(run_gaussian_mixture(fit_gaussian_mixture)[0])
plotout(np.arange(len(proteins)), ks, ks2)
del_list = []
for res, ddG_res in enumerate(ddGs):
ddGs_17 = fit_gaussian_mixture.preprocess(ddG_res)
if len(ddGs_17)!=17:
del_list.append(res)
return del_list
def plotout(ddGs_exp, fit_gaussian_mixture):
plt.title('$\\beta$1 domain of protein G')
plt.hist(ddGs_exp[~np.isnan(ddGs_exp)], fit_gaussian_mixture.bins, density = True)
plt.plot(fit_gaussian_mixture.x, yPred, color='k', label = 'N-Gaussian\nD = {0:.2f}\np-value = {1:.2E}'.format(k, KS_pvalue(k, len(fit_gaussian_mixture.x), len(fit_gaussian_mixture.x))))
plt.plot(fit_gaussian_mixture.x, gaussian, color='r', label = 'bi-Gaussian\nD = {0:.2f}\np-value = {1:.2E}'.format(ks2, KS_pvalue(ks2, len(fit_gaussian_mixture.x), len(fit_gaussian_mixture.x))))
plt.xlabel('$\Delta \Delta G$ (kcal/mol)')
plt.ylabel('probability')
plt.legend()
plt.show()
if '__main__':
ddGs_exp = read_in_experiment()
fit_gaussian_mixture = FitGaussianMixture(ddGs_exp)
del_list = parse(ddGs_exp)
ddGs_exp = np.delete(ddGs_exp, del_list, axis = 0)
yPred = generate_dist(ddGs_exp, fit_gaussian_mixture)
k, pvalue = stats.ks_2samp(fit_gaussian_mixture.hist, yPred)
ks2, pvalue2, gaussian = run_gaussian_mixture(fit_gaussian_mixture)
plotout(ddGs_exp, fit_gaussian_mixture)