Log In Sign Up
Models:
Marco Lenoir
MarcoTheBlack/
spatial-alignment
0
Downloads: 1
[54ded2]: / experiments / expression / slideseq / plot_prediction_results.py

Download this file

159 lines (120 with data), 5.0 kB 

  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
import matplotlib.pyplot as plt

import seaborn as sns

import pandas as pd

import numpy as np

from scipy.stats import pearsonr



import matplotlib



font = {"size": 30}

matplotlib.rc("font", **font)

matplotlib.rcParams["text.usetex"] = True





# error_df = pd.read_csv("./out/prediction_comparison_visium.csv", index_col=0)

# error_df = pd.read_csv("./out/twod_prediction_visium.csv", index_col=0)



errors_union = pd.read_csv("./out/prediction_errors_union.csv", index_col=0)

errors_separate = pd.read_csv("./out/prediction_errors_separate.csv", index_col=0)

errors_gpsa = pd.read_csv("./out/prediction_errors_gpsa.csv", index_col=0)



errors_union_melted = pd.melt(errors_union)

errors_union_melted["method"] = "Union"



errors_separate_melted = pd.melt(errors_separate)

errors_separate_melted["method"] = "Separate"



errors_gpsa_melted = pd.melt(errors_gpsa)

errors_gpsa_melted["method"] = "GPSA"



results_df = pd.concat(

    [errors_union_melted, errors_separate_melted, errors_gpsa_melted], axis=0

)





results_df_means = results_df.groupby(["variable", "method"], as_index=False).mean()

results_df_stddevs = results_df.groupby(["variable", "method"], as_index=False).std()



results_df_gpsa = pd.merge(

    results_df_means[results_df_means.method == "GPSA"],

    results_df_stddevs[results_df_stddevs.method == "GPSA"],

    on=["variable", "method"],

    suffixes=["_mean", "_stddev"],

)

results_df_union = pd.merge(

    results_df_means[results_df_means.method == "Union"],

    results_df_stddevs[results_df_stddevs.method == "Union"],

    on=["variable", "method"],

    suffixes=["_mean", "_stddev"],

)

assert np.array_equal(results_df_gpsa.variable.values, results_df_union.variable.values)





plt.figure(figsize=(14, 7))



plt.subplot(121)



# results_df_trialwise_mean = results_df.groupby(["method", "variable"], as_index=False).mean()

# results_df_trialwise_mean = results_df_trialwise_mean[results_df_trialwise_mean.method != "Separate"]

results_df_trialwise_mean = pd.DataFrame(

    pd.concat([errors_union.mean(1), errors_gpsa.mean(1)]), columns=["value"]

)

results_df_trialwise_mean["method"] = np.concatenate(

    [["Union"] * len(errors_union), ["GPSA"] * len(errors_gpsa)]

)

g = sns.boxplot(data=results_df_trialwise_mean, x="method", y="value", color="gray")

plt.xlabel("")

plt.ylabel(r"Pearson $\rho$")

plt.suptitle("Slide-seqV2 prediction")





plt.subplot(122)

plt.errorbar(

    x=results_df_union.value_mean.values,

    y=results_df_gpsa.value_mean.values,

    xerr=results_df_union.value_stddev.values,

    yerr=results_df_gpsa.value_stddev.values,

    fmt="o",

    ecolor="black",

    color="black",

)



plt.xlabel(r"Pearson $\rho$, Union")

plt.ylabel(r"Pearson $\rho$, GPSA")



ax = plt.gca()



lims = [

    np.min([ax.get_xlim(), ax.get_ylim()]),  # min of both axes

    np.max([ax.get_xlim(), ax.get_ylim()]),  # max of both axes

]



# now plot both limits against eachother

ax.plot(lims, lims, "k-", alpha=0.75, zorder=0, color="gray")

ax.set_aspect("equal")

ax.set_xlim(lims)

ax.set_ylim(lims)

plt.tight_layout()



plt.savefig("./out/two_d_prediction_comparison_slideseq.png")

plt.show()





preds = pd.read_csv("./out/slideseq_preds_gpsa.csv", index_col=0)

truth = pd.read_csv("./out/slideseq_truth_gpsa.csv", index_col=0)



pearson_corrs = np.zeros(preds.shape[1])

for jj in range(preds.shape[1]):

    pearson_corrs[jj] = pearsonr(truth.iloc[:, jj].values, preds.iloc[:, jj].values)[0]



sorted_idx = np.argsort(pearson_corrs)

n_genes_to_plot = 3



plt.figure(figsize=(n_genes_to_plot * 7, 14))



gene_names = pd.read_csv("./out/slideseq_pred_gene_names.csv").iloc[:, 0].values



# import ipdb



# ipdb.set_trace()



for ii, gene_idx in enumerate(sorted_idx[-n_genes_to_plot:]):

    plt.subplot(2, n_genes_to_plot, ii + 1)

    plt.scatter(

        truth.iloc[:, gene_idx].values, preds.iloc[:, gene_idx].values, c="gray"

    )

    plt.xlabel("True expression")

    plt.ylabel("Predicted expression")

    plt.title(r"$\emph{" + gene_names[gene_idx].upper() + "}$")



for ii, gene_idx in enumerate(sorted_idx[:n_genes_to_plot]):

    plt.subplot(2, n_genes_to_plot, ii + 4)

    plt.scatter(

        truth.iloc[:, gene_idx].values, preds.iloc[:, gene_idx].values, c="gray"

    )

    plt.xlabel("True expression")

    plt.ylabel("Predicted expression")

    plt.title(r"$\emph{" + gene_names[gene_idx].upper() + "}$")



plt.tight_layout()

plt.savefig("./out/slideseq_prediction_examples.png")

plt.show()



# for jj in range(preds.shape[1]):

#     # import ipdb; ipdb.set_trace()

#     print(round(pearsonr(truth.iloc[:, jj].values, preds.iloc[:, jj].values)[0], 3))

#     nonzero_idx = np.where(truth.iloc[:, jj].values != np.min(truth.iloc[:, jj].values))[0]

#     print(round(pearsonr(truth.iloc[:, jj].values[nonzero_idx], preds.iloc[:, jj].values[nonzero_idx])[0], 3))

#     print()

#     plt.scatter(truth.iloc[:, jj].values, preds.iloc[:, jj].values, c="gray")

#     plt.xlabel("True expression")

#     plt.ylabel("Predicted expression")

#     plt.show()

#     # import ipdb; ipdb.set_trace()





import ipdb



ipdb.set_trace()