TCGA学习笔记6-主成份分析PCA

发表于 | 分类于 | | 热度 °C

上面，我们对RNA表达数据进行了差异分析，下面，我们将进行主成份分析（PCA），用于对数据进行区分和归类。

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
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
> pcaData <- data[rownames(result_select),] # 这里，首先使用差异表达数据，即筛选出来的4913个基因在611个样本中的表达水平
> class(pcaData)
[1] "data.frame"
> dim(pcaData)
[1] 4913  611
> head(pcaData)[,1:3]
                TCGA-CZ-5465-01A-01R-1503 TCGA-BP-4355-01A-01R-1289
ENSG00000000938                       899                      1211
ENSG00000001617                     16862                     10888
ENSG00000001630                       310                        77
ENSG00000002586                     10729                      8611
ENSG00000002746                         7                        37
ENSG00000002933                     60355                     29448
                TCGA-CZ-5451-01A-01R-1503
ENSG00000000938                      1100
ENSG00000001617                      9299
ENSG00000001630                       305
ENSG00000002586                      7618
ENSG00000002746                         7
ENSG00000002933                     47857
> pcaDataT <- as.data.frame(t(pcaData)) # 行列转置，每行一个样本，每列一个基因
> class(pcaDataT)
[1] "data.frame"
> dim(pcaDataT)
[1]  611 4913
> head(pcaDataT)[,1:3]
                          ENSG00000000938 ENSG00000001617 ENSG00000001630
TCGA-CZ-5465-01A-01R-1503             899           16862             310
TCGA-BP-4355-01A-01R-1289            1211           10888              77
TCGA-CZ-5451-01A-01R-1503            1100            9299             305
TCGA-B0-5081-01A-01R-1334            2065            9886             103
TCGA-CZ-5454-11A-01R-1503             783            7150             815
TCGA-B0-5697-01A-11R-1541            1857            5288             146
> pcaDataTGroup <- data.frame(pcaDataT,Group=group) # 将分类信息加上，方便后面作图时颜色来区分
> dim(pcaDataTGroup)
[1]  611 4914
> head(pcaDataTGroup)[,4912:4914]
                          ENSG00000281404 ENSG00000281490  Group
TCGA-CZ-5465-01A-01R-1503              73             446 cancer
TCGA-BP-4355-01A-01R-1289             209            1208 cancer
TCGA-CZ-5451-01A-01R-1503              17             126 cancer
TCGA-B0-5081-01A-01R-1334             180             626 cancer
TCGA-CZ-5454-11A-01R-1503              36             240 normal
TCGA-B0-5697-01A-11R-1541              42             232 cancer

> pca <- prcomp(pcaDataT,scale=TRUE) # 使用R自带的prcomp()函数
> summary(pca)
Importance of components:
                           PC1      PC2      PC3      PC4      PC5      PC6      PC7
Standard deviation     30.5406 21.06796 17.82409 15.56497 13.67455 11.96656 11.39468
                            PC8     PC9    PC10    PC11    PC12    PC13    PC14    PC15
Standard deviation     10.52719 9.12744 8.76353 7.25503 6.82239 6.62584 6.35338 6.16851
                          PC16    PC17    PC18    PC19    PC20    PC21    PC22    PC23
Standard deviation     6.10537 5.99082 5.87263 5.73438 5.61851 5.53551 5.40397 5.20705
                          PC24    PC25    PC26    PC27    PC28    PC29    PC30    PC31
Standard deviation     5.07502 4.94755 4.77045 4.73712 4.66466 4.60046 4.50403 4.48497
                          PC32    PC33    PC34    PC35    PC36    PC37    PC38    PC39
Standard deviation     4.41300 4.40078 4.24311 4.21938 4.11894 4.10427 4.03498 4.00425
                          PC40    PC41    PC42    PC43    PC44    PC45    PC46    PC47
Standard deviation     3.95901 3.92036 3.85311 3.81686 3.76335 3.72471 3.67439 3.65233
                          PC48    PC49    PC50    PC51    PC52    PC53    PC54    PC55
Standard deviation     3.62492 3.58215 3.53086 3.51850 3.49220 3.46807 3.40228 3.38611
                          PC56    PC57    PC58   PC59    PC60    PC61    PC62    PC63
Standard deviation     3.34341 3.33820 3.32907 3.2906 3.26280 3.23725 3.22031 3.19124
                          PC64    PC65    PC66    PC67    PC68    PC69    PC70    PC71
Standard deviation     3.14392 3.10767 3.09295 3.07358 3.02365 3.01153 2.99903 2.98530
                          PC72    PC73    PC74    PC75    PC76   PC77    PC78    PC79
Standard deviation     2.96770 2.95324 2.94226 2.92450 2.91141 2.8906 2.87615 2.84936
                          PC80    PC81    PC82    PC83    PC84    PC85    PC86    PC87
Standard deviation     2.82099 2.78651 2.78489 2.77050 2.76802 2.74951 2.73087 2.72602
                          PC88    PC89    PC90    PC91    PC92   PC93    PC94    PC95
Standard deviation     2.70287 2.69798 2.66500 2.65009 2.62908 2.6260 2.61238 2.59281
                          PC96    PC97    PC98    PC99   PC100  PC101   PC102   PC103
Standard deviation     2.59100 2.57340 2.55959 2.54633 2.53601 2.5261 2.50279 2.48348
                         PC104   PC105   PC106  PC107   PC108   PC109   PC110   PC111
Standard deviation     2.47543 2.45592 2.45236 2.4317 2.41468 2.41164 2.38742 2.37613
                         PC112   PC113   PC114  PC115  PC116   PC117   PC118   PC119   PC120
Standard deviation     2.35853 2.35056 2.33463 2.3295 2.3250 2.30730 2.30597 2.29750 2.29033
                         PC121   PC122   PC123   PC124   PC125  PC126   PC127   PC128
Standard deviation     2.26585 2.25312 2.24279 2.23134 2.22793 2.2158 2.20129 2.19728
                         PC129   PC130   PC131   PC132   PC133   PC134   PC135  PC136  PC137
Standard deviation     2.19296 2.17398 2.16039 2.14977 2.14574 2.13250 2.11185 2.1077 2.1031
                         PC138   PC139   PC140   PC141   PC142   PC143   PC144   PC145
Standard deviation     2.08261 2.07534 2.07042 2.05222 2.04793 2.04577 2.03694 2.02771
                         PC146   PC147   PC148   PC149  PC150  PC151   PC152   PC153   PC154
Standard deviation     2.01782 2.00092 1.99539 1.98983 1.9861 1.9781 1.96918 1.95985 1.95552
                         PC155   PC156   PC157   PC158   PC159   PC160   PC161   PC162
Standard deviation     1.94804 1.94291 1.93641 1.92242 1.92021 1.90568 1.89667 1.89451
                         PC163   PC164   PC165   PC166   PC167  PC168  PC169   PC170   PC171
Standard deviation     1.88673 1.88252 1.87848 1.87202 1.86426 1.8584 1.8548 1.84692 1.84509
                         PC172   PC173   PC174   PC175   PC176   PC177   PC178   PC179
Standard deviation     1.83931 1.83135 1.82669 1.82031 1.81080 1.80627 1.80070 1.79712
                         PC180   PC181   PC182   PC183   PC184   PC185   PC186   PC187
Standard deviation     1.78550 1.77936 1.77550 1.76623 1.75981 1.75181 1.74749 1.74488
                         PC188   PC189  PC190   PC191   PC192   PC193   PC194   PC195
Standard deviation     1.73577 1.73361 1.7181 1.70630 1.70290 1.69972 1.69536 1.68780
                         PC196   PC197   PC198   PC199   PC200   PC201   PC202   PC203
Standard deviation     1.67993 1.67835 1.67086 1.66480 1.66048 1.65500 1.65183 1.64799
                         PC204   PC205   PC206   PC207   PC208   PC209   PC210   PC211
Standard deviation     1.64225 1.64076 1.63780 1.63251 1.62520 1.61368 1.61132 1.60909
                         PC212   PC213   PC214   PC215   PC216   PC217  PC218  PC219  PC220
Standard deviation     1.60139 1.59764 1.59602 1.59042 1.58228 1.57580 1.5733 1.5654 1.5610
                         PC221   PC222   PC223   PC224   PC225   PC226   PC227   PC228
Standard deviation     1.55563 1.55465 1.54655 1.54196 1.53549 1.53234 1.52645 1.52422
                         PC229   PC230   PC231   PC232   PC233   PC234   PC235   PC236
Standard deviation     1.51790 1.51457 1.51286 1.50343 1.50315 1.49795 1.49454 1.48776
                         PC237   PC238   PC239   PC240   PC241   PC242   PC243   PC244
Standard deviation     1.48438 1.47875 1.47257 1.46959 1.46717 1.45983 1.45757 1.45308
                         PC245   PC246   PC247   PC248   PC249   PC250   PC251   PC252
Standard deviation     1.44916 1.44823 1.44553 1.44368 1.43596 1.42866 1.42567 1.42135
                         PC253   PC254   PC255  PC256  PC257  PC258  PC259   PC260   PC261
Standard deviation     1.41682 1.41351 1.41118 1.4086 1.4026 1.3972 1.3949 1.38863 1.38630
                         PC262   PC263   PC264   PC265   PC266   PC267   PC268   PC269
Standard deviation     1.38335 1.37999 1.37650 1.37295 1.36460 1.36280 1.35885 1.35620
                         PC270   PC271   PC272   PC273   PC274   PC275   PC276   PC277
Standard deviation     1.34918 1.34286 1.34120 1.33772 1.33425 1.32969 1.32843 1.32246
                         PC278   PC279   PC280   PC281   PC282   PC283   PC284   PC285
Standard deviation     1.32087 1.32038 1.31198 1.30642 1.30391 1.30088 1.29874 1.29487
                         PC286   PC287   PC288   PC289   PC290   PC291   PC292   PC293
Standard deviation     1.29066 1.28536 1.28144 1.27949 1.27251 1.27007 1.26710 1.26414
                         PC294   PC295   PC296   PC297   PC298   PC299   PC300   PC301
Standard deviation     1.26220 1.26014 1.25852 1.25420 1.25096 1.24993 1.24498 1.23962
                         PC302   PC303   PC304  PC305  PC306  PC307  PC308  PC309  PC310
Standard deviation     1.23400 1.23255 1.22667 1.2240 1.2231 1.2229 1.2214 1.2143 1.2102
                        PC311  PC312   PC313   PC314   PC315   PC316   PC317   PC318   PC319
Standard deviation     1.2100 1.2048 1.19990 1.19701 1.19515 1.18987 1.18491 1.18245 1.18014
                         PC320   PC321   PC322   PC323   PC324   PC325   PC326   PC327
Standard deviation     1.17643 1.17621 1.17231 1.16879 1.16163 1.16122 1.15969 1.15713
                         PC328   PC329   PC330   PC331   PC332   PC333   PC334   PC335
Standard deviation     1.15556 1.15240 1.14952 1.14577 1.14304 1.14126 1.13763 1.13693
                         PC336   PC337   PC338   PC339   PC340   PC341   PC342   PC343
Standard deviation     1.13375 1.13120 1.12788 1.12463 1.12139 1.11960 1.11934 1.11628
                         PC344   PC345   PC346   PC347   PC348   PC349   PC350   PC351
Standard deviation     1.11177 1.10918 1.10520 1.10440 1.10241 1.09819 1.09693 1.09243
                         PC352   PC353   PC354   PC355   PC356   PC357   PC358   PC359
Standard deviation     1.08730 1.08648 1.08409 1.08042 1.07850 1.07407 1.07129 1.06942
                         PC360   PC361   PC362   PC363   PC364   PC365   PC366   PC367
Standard deviation     1.06691 1.06346 1.06228 1.05964 1.05558 1.05401 1.05002 1.04714
                         PC368   PC369   PC370   PC371   PC372   PC373   PC374   PC375
Standard deviation     1.04667 1.04370 1.04314 1.03839 1.03722 1.03520 1.02846 1.02695
                         PC376   PC377   PC378   PC379   PC380   PC381   PC382   PC383
Standard deviation     1.02537 1.02305 1.02137 1.02052 1.01721 1.01152 1.01111 1.00876
                         PC384   PC385  PC386  PC387  PC388  PC389  PC390  PC391  PC392
Standard deviation     1.00659 1.00372 1.0015 1.0003 0.9959 0.9945 0.9933 0.9895 0.9853
                        PC393  PC394   PC395   PC396   PC397   PC398   PC399   PC400   PC401
Standard deviation     0.9813 0.9801 0.97715 0.97527 0.97116 0.96965 0.96598 0.96351 0.96042
                         PC402   PC403   PC404   PC405   PC406   PC407   PC408   PC409
Standard deviation     0.95959 0.95777 0.95501 0.95072 0.94820 0.94598 0.94302 0.94201
                         PC410   PC411   PC412   PC413   PC414   PC415   PC416   PC417
Standard deviation     0.93933 0.93682 0.93610 0.93316 0.93053 0.92868 0.92602 0.92544
                         PC418   PC419   PC420   PC421   PC422   PC423   PC424   PC425
Standard deviation     0.91855 0.91759 0.91581 0.91240 0.91112 0.91077 0.90678 0.90618
                         PC426   PC427   PC428   PC429   PC430   PC431   PC432   PC433
Standard deviation     0.90249 0.89949 0.89669 0.89384 0.89288 0.88850 0.88718 0.88584
                         PC434   PC435   PC436   PC437   PC438   PC439   PC440   PC441
Standard deviation     0.88341 0.87924 0.87845 0.87462 0.87209 0.86974 0.86603 0.86518
                         PC442   PC443   PC444   PC445   PC446   PC447   PC448   PC449
Standard deviation     0.86083 0.86061 0.85783 0.85555 0.85338 0.85200 0.85090 0.84969
                         PC450   PC451   PC452   PC453   PC454   PC455   PC456   PC457
Standard deviation     0.84627 0.84291 0.84245 0.83924 0.83537 0.83425 0.83233 0.82828
                         PC458   PC459   PC460   PC461   PC462   PC463   PC464   PC465
Standard deviation     0.82678 0.82414 0.82309 0.82012 0.81847 0.81767 0.81494 0.81351
                         PC466   PC467   PC468   PC469   PC470   PC471   PC472   PC473
Standard deviation     0.80988 0.80492 0.80108 0.80043 0.79918 0.79730 0.79388 0.79141
                         PC474   PC475   PC476   PC477   PC478   PC479   PC480   PC481
Standard deviation     0.78887 0.78738 0.78554 0.78497 0.78066 0.78024 0.77470 0.77313
                         PC482   PC483   PC484   PC485   PC486   PC487   PC488   PC489
Standard deviation     0.77203 0.77000 0.76624 0.76326 0.76087 0.75905 0.75893 0.75472
                         PC490   PC491   PC492   PC493   PC494   PC495   PC496   PC497
Standard deviation     0.75419 0.75251 0.74832 0.74672 0.74510 0.74337 0.74174 0.74037
                         PC498   PC499   PC500   PC501   PC502   PC503   PC504   PC505
Standard deviation     0.73802 0.73601 0.73324 0.73171 0.72897 0.72684 0.72589 0.72385
                         PC506  PC507  PC508  PC509  PC510  PC511  PC512  PC513  PC514
Standard deviation     0.72215 0.7172 0.7142 0.7132 0.7091 0.7087 0.7080 0.7049 0.7013
                        PC515  PC516  PC517  PC518  PC519  PC520  PC521   PC522   PC523
Standard deviation     0.6982 0.6953 0.6942 0.6924 0.6904 0.6880 0.6864 0.68305 0.68213
                         PC524   PC525   PC526   PC527   PC528   PC529   PC530   PC531
Standard deviation     0.67959 0.67589 0.67421 0.67178 0.66804 0.66633 0.66601 0.66377
                         PC532   PC533   PC534   PC535   PC536   PC537   PC538   PC539
Standard deviation     0.66191 0.65937 0.65725 0.65584 0.65419 0.65095 0.64724 0.64534
                         PC540   PC541   PC542   PC543   PC544   PC545   PC546   PC547
Standard deviation     0.64276 0.63951 0.63786 0.63523 0.63117 0.62784 0.62692 0.62319
                         PC548   PC549   PC550   PC551   PC552   PC553   PC554   PC555
Standard deviation     0.62051 0.61775 0.61483 0.61238 0.61000 0.60674 0.60389 0.60303
                         PC556   PC557   PC558   PC559   PC560   PC561   PC562   PC563
Standard deviation     0.60151 0.60121 0.59484 0.59220 0.58997 0.58819 0.58589 0.58175
                         PC564   PC565   PC566   PC567   PC568   PC569   PC570   PC571
Standard deviation     0.57990 0.57774 0.57318 0.57104 0.56929 0.56368 0.56112 0.55675
                         PC572   PC573   PC574   PC575   PC576   PC577   PC578   PC579
Standard deviation     0.55548 0.55188 0.55145 0.54780 0.54517 0.53832 0.53484 0.53255
                         PC580   PC581   PC582   PC583   PC584   PC585   PC586   PC587
Standard deviation     0.52787 0.52379 0.52309 0.51867 0.51252 0.51120 0.50921 0.50802
                         PC588   PC589   PC590   PC591   PC592   PC593   PC594   PC595
Standard deviation     0.50304 0.49706 0.49622 0.48912 0.48581 0.48163 0.48029 0.47163
                         PC596   PC597   PC598   PC599   PC600   PC601   PC602   PC603
Standard deviation     0.46913 0.46281 0.45966 0.45449 0.45346 0.44524 0.44136 0.43279
                         PC604   PC605   PC606   PC607   PC608   PC609   PC610     PC611
Standard deviation     0.41556 0.39920 0.39283 0.38120 0.37238 0.35365 0.31898 6.713e-15
 [ reached getOption("max.print") -- omitted 2 rows ]

> plot(pca,type="l") # 作图，看一下各个PC的权重，可见，PC1和PC2占了绝大多数权重

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
> str(pca) # 看一下数据结构
List of 5
 $ sdev    : num [1:611] 30.5 21.1 17.8 15.6 13.7 ...
 $ rotation: num [1:4913, 1:611] -0.0177 -0.0113 0.0172 -0.0142 0.0154 ...
  ..- attr(*, "dimnames")=List of 2
  .. ..$ : chr [1:4913] "ENSG00000000938" "ENSG00000001617" "ENSG00000001630" "ENSG00000002586" ...
  .. ..$ : chr [1:611] "PC1" "PC2" "PC3" "PC4" ...
 $ center  : Named num [1:4913] 1086 6694 235 9934 241 ...
  ..- attr(*, "names")= chr [1:4913] "ENSG00000000938" "ENSG00000001617" "ENSG00000001630" "ENSG00000002586" ...
 $ scale   : Named num [1:4913] 743 3689 183 5885 619 ...
  ..- attr(*, "names")= chr [1:4913] "ENSG00000000938" "ENSG00000001617" "ENSG00000001630" "ENSG00000002586" ...
 $ x       : num [1:611, 1:611] -19.09 -35.24 9.44 -58.72 91.79 ...
  ..- attr(*, "dimnames")=List of 2
  .. ..$ : chr [1:611] "TCGA-CZ-5465-01A-01R-1503" "TCGA-BP-4355-01A-01R-1289" "TCGA-CZ-5451-01A-01R-1503" "TCGA-B0-5081-01A-01R-1334" ...
  .. ..$ : chr [1:611] "PC1" "PC2" "PC3" "PC4" ...
 - attr(*, "class")= chr "prcomp"

> data_pca <- cbind(pcaDataTGroup,pca$x[,1:2]) # 将PC1与PC2加进去
> dim(data_pca)
[1]  611 4916
> head(data_pca)[,4913:4916]
                          ENSG00000281490  Group        PC1        PC2
TCGA-CZ-5465-01A-01R-1503             446 cancer -19.085520  10.325704
TCGA-BP-4355-01A-01R-1289            1208 cancer -35.244522  34.570355
TCGA-CZ-5451-01A-01R-1503             126 cancer   9.439234 -23.982777
TCGA-B0-5081-01A-01R-1334             626 cancer -58.723514  23.949627
TCGA-CZ-5454-11A-01R-1503             240 normal  91.788022  78.416490
TCGA-B0-5697-01A-11R-1541             232 cancer  -1.465493  -2.567302

# PLOT WITH GGPLOT -----
> library(ggplot2)
> ggplot(data_pca,aes(PC1,PC2,col=Group,fill=Group))+
+   stat_ellipse(geom="polygon",col="black",alpha=0.5)+
+   geom_point(shape=21,col="black",size=1.2)+
+   theme(panel.background = element_rect(fill="transparent",color="black"),
+         panel.grid.minor = element_blank(), 
+         panel.grid.major = element_blank())

如图，肿瘤样本与正常样本被很好地分为两组，可以明确区分，这也可被用于预测一个未知样本是否是肿瘤样本或是正常样本。

  • 本文作者：括囊无誉
  • 本文链接： TCGA/TCGA6PCA/
  • 版权声明： 本博客所有文章均为原创作品，转载请注明出处！
------ 本文结束 ------
坚持原创文章分享，您的支持将鼓励我继续创作！