Chapter 10 Group Analysis and Visualization
Group analysis is a common task in cancer study. Sigminer supports dividing samples into multiple groups and comparing genotype/phenotype feature measures.
10.1 Group Generation
There are multiple methods to generate groups, including ‘consensus’ (default, can be only used by result from sig_extract()), ‘k-means’ etc. After determining groups, sigminer will assign each group to a signature with maximum fraction. We may say a group is Sig_x enriched.
mt_grps <- get_groups(mt_sig, method = "consensus", match_consensus = TRUE)
#> [36mℹ[39m [2020-10-09 00:03:46]: Started.
#> [32m✓[39m [2020-10-09 00:03:46]: 'Signature' object detected.
#> [36mℹ[39m [2020-10-09 00:03:46]: Obtaining clusters from the hierarchical clustering of the consensus matrix...
#> [36mℹ[39m [2020-10-09 00:03:46]: Finding the dominant signature of each group...
#> => Generating a table of group and dominant signature:
#>
#> Sig1 Sig2 Sig3 Sig4 Sig5
#> 1 14 0 0 0 0
#> 2 0 2 4 0 0
#> 3 1 0 8 15 0
#> 4 0 2 0 0 8
#> 5 0 42 3 1 0
#> => Assigning a group to a signature with the maxium fraction (stored in 'map_table' attr)...
#> [36mℹ[39m [2020-10-09 00:03:46]: Summarizing...
#> group #1: 14 samples with Sig1 enriched.
#> group #2: 6 samples with Sig3 enriched.
#> group #3: 24 samples with Sig4 enriched.
#> group #4: 10 samples with Sig5 enriched.
#> group #5: 46 samples with Sig2 enriched.
#> [33m![39m [2020-10-09 00:03:46]: The 'enrich_sig' column is set to dominant signature in one group, please check and make it consistent with biological meaning (correct it by hand if necessary).
#> [36mℹ[39m [2020-10-09 00:03:46]: 0.088 secs elapsed.
head(mt_grps)
#> sample group silhouette_width enrich_sig
#> 1: TCGA-PE-A5DD-01A-12D-A27P-09 1 0.572 Sig1
#> 2: TCGA-D8-A1JJ-01A-31D-A14K-09 1 0.279 Sig1
#> 3: TCGA-BH-A18K-01A-11D-A12B-09 1 0.125 Sig1
#> 4: TCGA-AC-A2FO-01A-11D-A17W-09 1 0.259 Sig1
#> 5: TCGA-E2-A1IH-01A-11D-A188-09 1 0.535 Sig1
#> 6: TCGA-E2-A152-01A-11D-A12B-09 1 0.482 Sig1The returned sample orders match sample orders in clustered consensus matrix.
Sometimes, the mapping between groups and enriched signatures may not right. Users should check it and even correct it manually.
10.2 Group Comparison Analysis
load(system.file("extdata", "toy_copynumber_signature_by_M.RData",
package = "sigminer", mustWork = TRUE
))
# Assign samples to clusters
groups <- get_groups(sig, method = "k-means")
#> [36mℹ[39m [2020-10-09 00:03:46]: Started.
#> [32m✓[39m [2020-10-09 00:03:46]: 'Signature' object detected.
#> [36mℹ[39m [2020-10-09 00:03:46]: Running k-means with 2 clusters...
#> [36mℹ[39m [2020-10-09 00:03:46]: Generating a table of group and signature contribution (stored in 'map_table' attr):
#> Sig1 Sig2
#> 1 0.003428449 0.9965716
#> 2 0.031799383 0.9682006
#> [36mℹ[39m [2020-10-09 00:03:46]: Assigning a group to a signature with the maximum fraction...
#> [36mℹ[39m [2020-10-09 00:03:46]: Summarizing...
#> group #1: 6 samples with Sig2 enriched.
#> group #2: 4 samples with Sig2 enriched.
#> [33m![39m [2020-10-09 00:03:46]: The 'enrich_sig' column is set to dominant signature in one group, please check and make it consistent with biological meaning (correct it by hand if necessary).
#> [36mℹ[39m [2020-10-09 00:03:46]: 0.042 secs elapsed.
set.seed(1234)
groups$prob <- rnorm(10)
groups$new_group <- sample(c("1", "2", "3", "4", NA), size = nrow(groups), replace = TRUE)
# Compare groups (filter NAs for categorical coloumns)
groups.cmp <- get_group_comparison(groups[, -1],
col_group = "group",
cols_to_compare = c("prob", "new_group"),
type = c("co", "ca"), verbose = TRUE
)
#> Treat prob as continuous variable.
#> Treat new_group as categorical variable.
# Compare groups (Set NAs of categorical columns to 'Rest')
groups.cmp2 <- get_group_comparison(groups[, -1],
col_group = "group",
cols_to_compare = c("prob", "new_group"),
type = c("co", "ca"), NAs = "Rest", verbose = TRUE
)
#> Treat prob as continuous variable.
#> Treat new_group as categorical variable.
