Analysis of TCGA RPPA LIHC samples

Last updated: 2021-10-01

Checks: 7 0

Knit directory: Multispectral HCC/

This reproducible R Markdown analysis was created with workflowr (version 1.6.2). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.

R Markdown file: up-to-date

Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.

Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

Seed: set.seed(20210728)

The command set.seed(20210728) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

Session information: recorded

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

Cache: none

Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.

File paths: relative

Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.

Repository version: d134fc2

Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.

The results in this page were generated with repository version d134fc2. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:


Ignored files:
    Ignored:    .DS_Store
    Ignored:    .Rhistory
    Ignored:    .Rproj.user/
    Ignored:    analysis/.DS_Store
    Ignored:    analysis/figure/
    Ignored:    code/
    Ignored:    data/
    Ignored:    old/
    Ignored:    output/.DS_Store
    Ignored:    output/tumor.hc.nmf.rank.1.rds
    Ignored:    output/tumor.hc.nmf.rank.10.rds
    Ignored:    output/tumor.hc.nmf.rank.5.rds
    Ignored:    output/tumor.hc.umap.rds

Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.

These are the previous versions of the repository in which changes were made to the R Markdown (analysis/tcga.Rmd) and HTML (docs/tcga.html) files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view the files as they were in that past version.

File	Version	Author	Date	Message
Rmd	13f7fb5	Jovan Tanevski	2021-10-01	leave only nmf
html	438a57b	Jovan Tanevski	2021-09-18	Build site.
Rmd	421ce3c	Jovan Tanevski	2021-09-18	add caching, bump cores to 10%
html	cfc1900	Jovan Tanevski	2021-09-17	Build site.
Rmd	410dc90	Jovan Tanevski	2021-09-17	fix issue with low number of examples
Rmd	201876f	Jovan Tanevski	2021-09-17	work with normalized data for profiles and de
html	c85e3e6	Jovan Tanevski	2021-09-17	Build site.
Rmd	b8745a8	Jovan Tanevski	2021-09-17	fix typos
html	ae61865	Jovan Tanevski	2021-09-16	Build site.
Rmd	9c11427	Jovan Tanevski	2021-09-16	fix difference calculation
html	9a18eda	Jovan Tanevski	2021-09-16	Build site.
Rmd	17fc915	Jovan Tanevski	2021-09-16	update hclust
Rmd	47928d9	Jovan Tanevski	2021-09-16	add silhouette based differential expression analysis, repeat nmf
html	e7709fe	Jovan Tanevski	2021-08-03	Build site.
html	f791e00	Jovan Tanevski	2021-08-03	Build site.
Rmd	e464694	Jovan Tanevski	2021-08-03	fix seed, stabilize nmf
html	51e2651	Jovan Tanevski	2021-08-03	Build site.
Rmd	7d62e48	Jovan Tanevski	2021-08-03	hclust and leiden on quantile normalized data
html	55fb7fc	Jovan Tanevski	2021-08-02	Build site.
Rmd	a152862	Jovan Tanevski	2021-08-02	add NMF based analysis
html	c25dcca	Jovan Tanevski	2021-08-02	Build site.
Rmd	21cf743	Jovan Tanevski	2021-08-02	rank normalization, leiden analysis, diff exp
html	a386eaa	Jovan Tanevski	2021-07-28	Build site.
Rmd	9ab4acc	Jovan Tanevski	2021-07-28	Build site.
Rmd	bd05dbf	Jovan Tanevski	2021-07-28	add expression profiles
html	e45bf1d	Jovan Tanevski	2021-07-28	Build site.
Rmd	dcab1bf	Jovan Tanevski	2021-07-28	set figure output to svg
html	33e71a0	Jovan Tanevski	2021-07-28	Build site.
Rmd	bb998b8	Jovan Tanevski	2021-07-28	hclust with factoextra
html	1cb80af	Jovan Tanevski	2021-07-28	Build site.
Rmd	d0b6e0f	Jovan Tanevski	2021-07-28	add pca to tcga analysis
html	d73f586	Jovan Tanevski	2021-07-28	Build site.
Rmd	c4324f5	Jovan Tanevski	2021-07-28	add umap
html	0bfae68	Jovan Tanevski	2021-07-28	Build site.
Rmd	82e3712	Jovan Tanevski	2021-07-28	add basic hclust

Setup

Load required libraries.

library(tidyverse)
library(skimr)
library(uwot)
library(factoextra)
library(cowplot)
library(limma)
library(NMF)
library(pheatmap)

Read filtered TCGA RRPA data and display summary statistics.

tcga.raw <- read_csv("data/TCGA-RPPA-LIHC_selected.csv", col_types = cols()) %>%
  select(-TumorType) %>%
  column_to_rownames("SampleID")

skim(tcga.raw)

Data summary
Name	tcga.raw
Number of rows	184
Number of columns	12
_______________________
Column type frequency:
numeric	12
________________________
Group variables	None

Variable type: numeric

skim_variable	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
P53	1	-1.13	0.46	-2.54	-1.40	-0.96	-0.76	-0.51	▁▂▂▅▇
AKT_pS473	1	-0.63	0.71	-2.76	-1.02	-0.41	-0.15	0.96	▁▂▃▇▁
AKT_pT308	1	0.07	0.42	-1.16	-0.07	0.09	0.22	1.72	▁▃▇▁▁
BETACATENIN	1	1.50	0.72	-1.30	1.08	1.51	1.87	3.70	▁▁▇▅▁
JNK_pT183Y185	1	-0.21	0.29	-1.14	-0.36	-0.17	-0.04	0.48	▁▂▇▇▂
MEK1_pS217S221	1	-0.17	0.36	-0.82	-0.35	-0.23	-0.10	2.58	▇▃▁▁▁
P38_pT180Y182	1	0.49	0.57	-1.45	0.31	0.50	0.72	2.95	▁▂▇▁▁
P70S6K_pT389	1	-1.21	0.75	-3.22	-1.58	-0.95	-0.65	1.09	▂▃▇▃▁
PDK1_pS241	1	0.40	0.31	-0.56	0.23	0.37	0.55	1.62	▁▇▇▂▁
S6_pS235S236	1	-0.72	0.75	-3.53	-1.06	-0.67	-0.36	1.12	▁▁▅▇▂
YAP_pS127	1	2.21	0.61	0.68	1.78	2.10	2.54	4.05	▁▇▇▂▁
TRANSGLUTAMINASE	1	-0.42	0.56	-1.24	-0.83	-0.59	-0.20	2.72	▇▃▁▁▁

Quantile normalization and rank normalization for NMF as suggested in https://gdac.broadinstitute.org/runs/analyses__2016_01_28/reports/cancer/LIHC/RPPA_Clustering_CNMF/nozzle.html

tcga.norm <- normalizeQuantiles(tcga.raw)

skim(tcga.norm)

Data summary
Name	tcga.norm
Number of rows	184
Number of columns	12
_______________________
Column type frequency:
numeric	12
________________________
Group variables	None

Variable type: numeric

skim_variable	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
P53	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
AKT_pS473	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
AKT_pT308	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
BETACATENIN	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
JNK_pT183Y185	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
MEK1_pS217S221	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
P38_pT180Y182	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
P70S6K_pT389	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
PDK1_pS241	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
S6_pS235S236	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
YAP_pS127	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁
TRANSGLUTAMINASE	1	0.01	0.52	-1.59	-0.27	0.05	0.3	1.87	▁▃▇▂▁

tcga.rank <- mutate_all(tcga.raw, rank)

Dimensionality reduction

tcga.pca <- prcomp(tcga.norm)
summary(tcga.pca)

Importance of components:
                         PC1    PC2    PC3     PC4     PC5     PC6     PC7
Standard deviation     1.052 0.6800 0.5864 0.53531 0.50548 0.47429 0.40905
Proportion of Variance 0.340 0.1421 0.1057 0.08809 0.07855 0.06915 0.05144
Cumulative Proportion  0.340 0.4821 0.5878 0.67594 0.75449 0.82364 0.87508
                           PC8     PC9    PC10    PC11    PC12
Standard deviation     0.35923 0.34052 0.28251 0.23710 0.15913
Proportion of Variance 0.03967 0.03565 0.02454 0.01728 0.00778
Cumulative Proportion  0.91475 0.95040 0.97493 0.99222 1.00000

set.seed(42)
tcga.umap <- umap(tcga.norm, n_neighbors = 10, n_epochs = 1000)

Consensus NMF clustering

For this analysis 3 clusters were selected based on the consensus cophenetic correlation coefficient.

tcga.nmf.rank <- nmfEstimateRank(as.matrix(t(tcga.rank)), seq(2,10), nrun = 20, seed = 42)

plot(tcga.nmf.rank)

tcga.nmf <- tcga.nmf <- tcga.nmf.rank$fit[["3"]]

Extract basis of NMF (signature of cluster)

basismap(tcga.nmf)

Extract coefficients of NMF (soft clustering of samples)

coefmap(tcga.nmf)

Check for signs of overfitting

consensusmap(tcga.nmf)

Assign clusters

nmf.clusters <- apply(tcga.nmf@fit@H, 2, which.max)

Plot in 2D PCA and UMAP

fviz_pca_ind(tcga.pca, geom = "point", col.ind = as.factor(nmf.clusters)) +
  theme_classic()

tcga.umap.clus <-
  tcga.umap %>%
  cbind(nmf.clusters) %>%
  `colnames<-`(c("U1", "U2", "Cluster")) %>%
  as_tibble() %>%
  mutate_at("Cluster", as.factor)

ggplot(tcga.umap.clus, aes(x = U1, y = U2, color = Cluster)) +
  geom_point() +
  theme_classic()

Expression profiles per cluster

tcga.clustered.nmf <- tcga.norm %>%
  mutate(Cluster = as.factor(nmf.clusters)) %>%
  pivot_longer(names_to = "Marker", values_to = "Norm.value", -Cluster)

profiles <- seq_len(max(nmf.clusters)) %>% map(~
ggplot(
  tcga.clustered.nmf %>% filter(Cluster == .x),
  aes(x = Marker, y = Norm.value, color = Marker)
) +
  stat_summary(fun.data = mean_sdl, show.legend = FALSE) +
  ylim(-3, 3) +
  theme_classic() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1)))

plot_grid(plotlist = profiles, labels = paste("Cluster", seq_len(max(nmf.clusters))))

Differential expression analysis (limma)

design <- model.matrix(~0 + as.factor(nmf.clusters))
colnames(design) <- paste0("Cluster", seq_len(max(nmf.clusters)))

tcga.nmf.limma <- lmFit(t(tcga.norm), design = design)

tcga.nmf.eb <- eBayes(tcga.nmf.limma)

topTable(tcga.nmf.eb)

	Cluster1	Cluster2	Cluster3	AveExpr	F	P.Value	adj.P.Val
P53	-0.4414233	-0.4485783	0.2956533	0.0140382	49.62915	0e+00	0e+00
AKT_pS473	-0.5334368	-0.3138352	0.2904909	0.0140382	49.52718	0e+00	0e+00
P70S6K_pT389	-0.3765563	-0.4171258	0.2649090	0.0140382	35.36002	0e+00	0e+00
MEK1_pS217S221	0.5545074	-0.3643842	-0.0679547	0.0140382	29.63664	0e+00	0e+00
JNK_pT183Y185	-0.0595382	-0.5230228	0.1852520	0.0140382	20.60883	0e+00	0e+00
BETACATENIN	-0.3085174	0.5078866	-0.0099059	0.0140382	18.98657	0e+00	0e+00
PDK1_pS241	0.2734628	0.4012811	-0.1799376	0.0140863	18.75428	0e+00	0e+00
S6_pS235S236	0.0790752	-0.5132934	0.1351859	0.0140382	16.62621	0e+00	0e+00
AKT_pT308	-0.2971225	-0.2369574	0.1887412	0.0140382	14.32245	0e+00	0e+00
YAP_pS127	0.0589849	0.4566955	-0.1217100	0.0140382	12.63077	1e-07	1e-07

tests.nmf <- decideTests(tcga.nmf.eb)

tests.nmf@.Data

                 Cluster1 Cluster2 Cluster3
P53                    -1       -1        1
AKT_pS473              -1       -1        1
AKT_pT308              -1       -1        1
BETACATENIN            -1        1        0
JNK_pT183Y185           0       -1        1
MEK1_pS217S221          1       -1        0
P38_pT180Y182           0       -1        1
P70S6K_pT389           -1       -1        1
PDK1_pS241              1        1       -1
S6_pS235S236            0       -1        1
YAP_pS127               0        1       -1
TRANSGLUTAMINASE        1        0        0

summary(tests.nmf)

       Cluster1 Cluster2 Cluster3
Down          5        8        2
NotSig        4        1        3
Up            3        3        7

Differential expression analysis (silhouette)

Calculate the similarity of samples using the expression and the silhouette scores based on the assigned clusters.

silhouette.nmf <- silhouette(nmf.clusters, dist(tcga.norm))
fviz_silhouette(silhouette.nmf)

  cluster size ave.sil.width
1       1   39         -0.04
2       2   31          0.15
3       3  114          0.33

Select only the samples with positive silhouette scores as “core samples”

core.samples <- which(silhouette.nmf[,3] > 0)
tcga.core.samples <- tcga.norm %>% add_column(Cluster = nmf.clusters) %>% 
  slice(core.samples)

Calculate difference in means (mean(cluster) - mean(other)), one-vs-all t-test per marker and correct for FDR. Filter q <= 0.05. Plot the differences.

de.table <- unique(tcga.core.samples$Cluster) %>% map_dfr(\(c){
  tcga.core.samples %>% 
    summarize(across(-Cluster, ~t.test(.x ~ (Cluster==c))$p.value)) %>%
    pivot_longer(names_to = "Marker", values_to = "p", everything()) %>% 
    mutate(Cluster = c, Difference = tcga.core.samples %>% 
             group_by(Cluster == c) %>% 
             select(-Cluster) %>% 
             group_split(.keep = FALSE) %>% map(colMeans) %>% reduce(`-`))
 }) %>% 
  mutate(q = p.adjust(p, method = "fdr"), Difference = -Difference)

de.table %>%
  filter(q <= 0.05) %>%
  arrange(q)

Marker	p	Cluster	Difference	q
P53	0.0000000	3	0.8261853	0.0000000
P70S6K_pT389	0.0000000	3	0.7636210	0.0000000
AKT_pS473	0.0000000	3	0.8341263	0.0000000
JNK_pT183Y185	0.0000000	2	-0.7426361	0.0000000
P53	0.0000000	2	-0.7245959	0.0000000
AKT_pS473	0.0000000	2	-0.6138768	0.0000000
P70S6K_pT389	0.0000000	2	-0.6583089	0.0000000
S6_pS235S236	0.0000000	2	-0.7278831	0.0000000
BETACATENIN	0.0000000	2	0.6267117	0.0000000
PDK1_pS241	0.0000001	2	0.5861173	0.0000004
MEK1_pS217S221	0.0000007	1	0.9512471	0.0000022
P38_pT180Y182	0.0000011	3	0.5284230	0.0000032
AKT_pT308	0.0000020	3	0.5371256	0.0000054
P38_pT180Y182	0.0000027	2	-0.5664680	0.0000069
JNK_pT183Y185	0.0000049	3	0.5255499	0.0000118
PDK1_pS241	0.0000081	3	-0.5210833	0.0000181
AKT_pS473	0.0000109	1	-1.0376493	0.0000230
P70S6K_pT389	0.0000241	1	-0.7212306	0.0000482
YAP_pS127	0.0000988	3	-0.5056444	0.0001872
P53	0.0001748	1	-0.7509946	0.0003146
YAP_pS127	0.0002097	2	0.5573314	0.0003595
AKT_pT308	0.0003973	2	-0.4833504	0.0006247
BETACATENIN	0.0003991	1	-0.8469248	0.0006247
S6_pS235S236	0.0004766	3	0.4551491	0.0007149
MEK1_pS217S221	0.0007290	2	-0.3592819	0.0010497
TRANSGLUTAMINASE	0.0072461	1	0.6227038	0.0100330
AKT_pT308	0.0122702	1	-0.4591054	0.0163602

de.table %>%
  pivot_wider(names_from = "Cluster", values_from = "Difference", -c(p, q)) %>%
  column_to_rownames("Marker") %>%
  as.matrix() %>%
  pheatmap(scale = "none")

sessionInfo()

R version 4.1.1 (2021-08-10)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS Big Sur 10.16

Matrix products: default
BLAS:   /Library/Frameworks/R.framework/Versions/4.1/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.1/Resources/lib/libRlapack.dylib

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] parallel  stats     graphics  grDevices utils     datasets  methods  
[8] base     

other attached packages:
 [1] RColorBrewer_1.1-2  doParallel_1.0.16   iterators_1.0.13   
 [4] foreach_1.5.1       pheatmap_1.0.12     NMF_0.23.0         
 [7] synchronicity_1.3.5 bigmemory_4.5.36    Biobase_2.52.0     
[10] BiocGenerics_0.38.0 cluster_2.1.2       rngtools_1.5.2     
[13] pkgmaker_0.32.2     registry_0.5-1      limma_3.48.3       
[16] cowplot_1.1.1       factoextra_1.0.7    uwot_0.1.10        
[19] Matrix_1.3-4        skimr_2.1.3         forcats_0.5.1      
[22] stringr_1.4.0       dplyr_1.0.7         purrr_0.3.4        
[25] readr_2.0.2         tidyr_1.1.4         tibble_3.1.4       
[28] ggplot2_3.3.5       tidyverse_1.3.1     workflowr_1.6.2    

loaded via a namespace (and not attached):
  [1] bigmemory.sri_0.1.3 colorspace_2.0-2    ggsignif_0.6.3     
  [4] rio_0.5.27          ellipsis_0.3.2      rprojroot_2.0.2    
  [7] htmlTable_2.2.1     base64enc_0.1-3     fs_1.5.0           
 [10] rstudioapi_0.13     ggpubr_0.4.0        farver_2.1.0       
 [13] ggrepel_0.9.1       bit64_4.0.5         RSpectra_0.16-0    
 [16] fansi_0.5.0         lubridate_1.7.10    xml2_1.3.2         
 [19] splines_4.1.1       codetools_0.2-18    knitr_1.36         
 [22] Formula_1.2-4       jsonlite_1.7.2      broom_0.7.9        
 [25] gridBase_0.4-7      dbplyr_2.1.1        png_0.1-7          
 [28] compiler_4.1.1      httr_1.4.2          backports_1.2.1    
 [31] assertthat_0.2.1    fastmap_1.1.0       cli_3.0.1          
 [34] later_1.3.0         htmltools_0.5.2     tools_4.1.1        
 [37] gtable_0.3.0        glue_1.4.2          reshape2_1.4.4     
 [40] Rcpp_1.0.7          carData_3.0-4       cellranger_1.1.0   
 [43] jquerylib_0.1.4     vctrs_0.3.8         xfun_0.26          
 [46] openxlsx_4.2.4      rvest_1.0.1         lifecycle_1.0.1    
 [49] rstatix_0.7.0       scales_1.1.1        vroom_1.5.5        
 [52] hms_1.1.1           promises_1.2.0.1    curl_4.3.2         
 [55] yaml_2.2.1          gridExtra_2.3       sass_0.4.0         
 [58] rpart_4.1-15        latticeExtra_0.6-29 stringi_1.7.4      
 [61] highr_0.9           checkmate_2.0.0     zip_2.2.0          
 [64] repr_1.1.3          rlang_0.4.11        pkgconfig_2.0.3    
 [67] evaluate_0.14       lattice_0.20-45     htmlwidgets_1.5.4  
 [70] labeling_0.4.2      bit_4.0.4           tidyselect_1.1.1   
 [73] plyr_1.8.6          magrittr_2.0.1      R6_2.5.1           
 [76] Hmisc_4.5-0         generics_0.1.0      DBI_1.1.1          
 [79] foreign_0.8-81      pillar_1.6.3        haven_2.4.3        
 [82] whisker_0.4         withr_2.4.2         nnet_7.3-16        
 [85] survival_3.2-13     abind_1.4-5         modelr_0.1.8       
 [88] crayon_1.4.1        car_3.0-11          uuid_0.1-4         
 [91] utf8_1.2.2          tzdb_0.1.2          rmarkdown_2.11     
 [94] jpeg_0.1-9          grid_4.1.1          readxl_1.3.1       
 [97] data.table_1.14.2   FNN_1.1.3           git2r_0.28.0       
[100] reprex_2.0.1        digest_0.6.28       xtable_1.8-4       
[103] httpuv_1.6.3        munsell_0.5.0       bslib_0.3.0

Analysis of TCGA RPPA LIHC samples

Jovan Tanevski

2021-10-01

Setup

Dimensionality reduction

Consensus NMF clustering