Code-for-MIT/genomicsLearnSmall.py at main · NGWi/Code-for-MIT · GitHub

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
from math import exp
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from sklearn.decomposition import PCA
from sklearn.manifold import TSNE, MDS
from sklearn.cluster import KMeans

data = np.load('../data/p1/X.npy')
print(data.shape)
print("Largest entry in first column is:", np.max(data[:, 0]))

pca = PCA(n_components=1)
explained_variance = pca.fit(data)
print("Variance explained by the first principal component is:", explained_variance.explained_variance_ratio_[0])


processed = np.log2(data + 1)
processed_variance = pca.fit(processed)
print("Largest entry in first column is:", np.max(processed[:, 0]))
print("Variance explained by the first principal component is:", processed_variance.explained_variance_ratio_[0])

plt.scatter(data[:, 0], data[:, 1])
plt.xlabel("First component")
plt.ylabel("Second component")
plt.title("Scatterplot of first two components of data")
plt.show()

# Change this to toggle between original and processed for the rest of the module
dataset = data

pca = PCA(n_components=2)
pca_projections = pca.fit_transform(dataset)

plt.scatter(pca_projections[:, 0], pca_projections[:, 1])
plt.xlabel("First principal component")
plt.ylabel("Second principal component")
plt.title("Scatterplot of first two principal components of data")
plt.show()

pca = PCA(n_components=50)
pca_projections = pca.fit_transform(dataset)

cum_variance = np.cumsum(pca.explained_variance_ratio_)
num_components = np.argmax(cum_variance >= 0.85) + 1
print("Number of PCs needed to explain 85% of the variance:", num_components)

best_kl_divergence = float('inf')
best_stress = float('inf')
pca = PCA(n_components=50)
pca_projections = pca.fit_transform(dataset)
kmeans = KMeans(n_clusters=10, random_state=np.random.RandomState())
cluster_assignments = kmeans.fit_predict(pca_projections)
for iteration in range(10):
    print("Iteration", iteration)
    tsne = TSNE(n_components=2, perplexity=40, random_state=np.random.RandomState())
    tsne_projections = tsne.fit_transform(pca_projections)
    kl_divergence = tsne.kl_divergence_
    if kl_divergence < best_kl_divergence:
        best_kl_divergence = kl_divergence
        best_tsne_projections = tsne_projections

    mds = MDS(n_components=2, random_state=np.random.RandomState())
    mds_projections = mds.fit_transform(dataset)
    stress = mds.stress_
    if stress < best_stress:
        best_stress = stress
        best_mds_projections = mds_projections

plt.scatter(pca_projections[:, 0], pca_projections[:, 1], c=cluster_assignments)
plt.xlabel("First dimension")
plt.ylabel("Second dimension")
plt.title("Scatterplot of first two dimensions of data using PCA")
plt.show()

plt.scatter(best_tsne_projections[:, 0], best_tsne_projections[:, 1], c=cluster_assignments)
plt.xlabel("First dimension")
plt.ylabel("Second dimension")
plt.title("Scatterplot of first two dimensions of data using PCA and T-SNE")
plt.show()

plt.scatter(best_mds_projections[:, 0], best_mds_projections[:, 1], c=cluster_assignments)
plt.xlabel("First dimension")
plt.ylabel("Second dimension")
plt.title("Scatterplot of first two dimensions of data using MDS")
plt.show()

inertias = []
for k in range(1, 11):
    kmeans = KMeans(n_clusters=k, random_state=np.random.RandomState())
    kmeans.fit(dataset)
    inertias.append(kmeans.inertia_)

print("Inertia for each k (WGSS):")
for k, inertia in enumerate(inertias):
    print(f"k={k+1}, Inertia={inertia:.3f}")
plt.plot(range(1, 11), inertias)
plt.xlabel("Number of clusters")
plt.ylabel("K-Means clustering criterion")
plt.title("Elbow method for selecting the number of clusters")
plt.show()

if len(dataset) <= 40:
    tsne_perplexity = len(dataset) - 1
else:
    tsne_perplexity = 40

kmeans = KMeans(n_clusters=10, random_state=np.random.RandomState())
cluster_assignments = kmeans.fit_predict(dataset)

cluster_means = np.array([dataset[cluster_assignments == i].mean(axis=0) for i in range(10)])

pca = PCA(n_components=2)
pca_projections = pca.fit_transform(cluster_means)

mds = MDS(n_components=2, random_state=np.random.RandomState())
mds_projections = mds.fit_transform(cluster_means)

plt.scatter(pca_projections[:, 0], pca_projections[:, 1])
plt.xlabel("First dimension")
plt.ylabel("Second dimension")
plt.title("Cluster means using PCA")
plt.show()

plt.scatter(mds_projections[:, 0], mds_projections[:, 1])
plt.xlabel("First dimension")
plt.ylabel("Second dimension")
plt.title("Cluster means using MDS")
plt.show()