tests/testthat/tests.R

In stemangiola/ARMET: Higher-Order Deconvolution Survival Analyses

library(dplyr)
library(ARMET)
data("test_mixture")
data("no_hierarchy_reference")

test_that("check data set",{
	
	# Test fo anylvel that all gnes are in all cel tpes
	expect_lte(
		ARMET::ARMET_ref |>
			distinct(level, symbol, `Cell type category`) |>
			count(level, symbol) |>
			distinct(level, n) |>
			nrow(),
		4
	)
	
	# Test dataset house keeping should be in all levels
	expect_equal(
		ARMET::ARMET_ref |> distinct(`house keeping`, level) |> nrow(),
		8
	)
	
	# Test dataset house keeping should be in all samples
	expect_equal(
		ARMET::ARMET_ref |> distinct(`house keeping`, sample) |> count(`house keeping`) |> distinct(n) |> nrow(),
		1
	)
	
	# there should be howse keeping
	expect_equal(
		ARMET::ARMET_ref |> distinct(`house keeping`) |> nrow(),
		2
	)
	
	# there should be howse keeping
	expect_equal(
		ARMET::ARMET_ref |> filter(level==3) |> pull(`Cell type category`) |> unique() |> intersect(c("endothelial", "epithelial", "fibroblast")) |> length(),
		3
	)
	
	# test if any count is NA
	expect_equal(
		ARMET::ARMET_ref |> filter(count |> is.na()) |> nrow(),
		0
	)
	
	
	
})

# my_mix =
# 	ARMET_ref |> 
# 	inner_join( (.) %>% distinct(sample) %>% slice(1:2)) %>% select(-level) %>%
# 	
# 	# complete
# 	mutate_if(is.factor, as.character) %>% 
# 	tidyr::complete(sample, symbol, fill = list(count = 0)) %>%
# 	mutate(count = as.integer(count))

test_that("check simple run NO hierarchy",{
	

	armet_estimate =
		# Read data
		test_mixture |>
		
		convoluted_glm(
			~ factor_of_interest,
			.sample = sample,
			.transcript = symbol,
			.abundance = count,
			reference = 
				readRDS("/wehisan/bioinf/bioinf-data/Papenfuss_lab/projects/mangiola.s/ARMET_dev/dev/TCGA_makeflow_pipeline/ref_jian_3_optimisations.rds") %>%
				filter(level==1),
			use_data = FALSE
		)
	
	
})


# test_that("check simple run",{
# 	
# 	armet_obj_hierarchical =
# 		
# 		# Read data
# 		my_mix |>
# 		nest(data = -sample) %>%
# 		mutate(factor_of_interest = c(0,1)) %>%
# 		unnest(data) %>%
# 		
# 		# Format
# 		ARMET:::setup_convolved_lm_hierarchical(
# 			~ factor_of_interest,
# 			.sample = sample,
# 			.transcript = symbol,
# 			.abundance = count,
# 			reference = readRDS("/wehisan/bioinf/bioinf-data/Papenfuss_lab/projects/mangiola.s/ARMET_dev/dev/TCGA_makeflow_pipeline/ref_jian_3_optimisations.rds")
# 		)
# 	
# 	armet_estimate =
# 		armet_obj_hierarchical |>
# 		estimate_convoluted_lm_1() 
# 	
# })

test_that("check nk dataset run",{
	
	cell_types = data.tree::as.Node(yaml::read_yaml("~/PostDoc/cellsig/dev/tree_kamran.yaml"))$leaves %>% purrr::map(~ .x$name) %>% as.character()
	
	genes = 
		readRDS("~/PostDoc/PPCG_tumour_microenvironment/methodwise_markers.rds") |> 
		filter(stream == "cibersortx") |> 
		pull(signature)
	
	cellsig_cibersortx_reference = 
		readRDS("~/PostDoc/cellsig/dev/modeling_results/counts_bayes_parsed.rds") |> 
		filter(cell_type %in% cell_types) |> 
		mutate(is_marker = .feature %in% genes) |> 
		select(symbol = .feature, cell_type, count= `50%`, is_marker) |> 
		distinct() 
	
		# Read data
	readRDS("~/PostDoc/cellsig/dev/counts.rds") |>
		filter(cell_type == "nk_resting") |> 
		tidyr::nest(data = -symbol) |> 
		mutate(n = purrr::map_int(data, ~ distinct(.x, sample) |> nrow())) |> 
		filter(n==max(n)) |> 
		select(-n) |> 
		tidyr::unnest(data) |> 

	#	filter(sample=="S02_B") |>  
		
		# Format
		convoluted_glm(
			~ 1,
			.sample = sample,
			.transcript = symbol,
			.abundance = count,
			reference = #cellsig_cibersortx_reference, 
				readRDS("/wehisan/bioinf/bioinf-data/Papenfuss_lab/projects/mangiola.s/ARMET_dev/dev/TCGA_makeflow_pipeline/ref_jian_3_optimisations.rds") %>%
				filter(level==3),
			use_cmdstanr = T
		) |>
		arrange(desc(`.median_(Intercept)`)) |> 
		pull(cell_type) |>
		magrittr::extract2(1) |> 
		as.character() |> 
		expect_equal("nk_resting")
	
	# # A tibble: 12 × 3
	# cell_type         `.median_(Intercept)` `.sd_(Intercept)`
	# <fct>                             <dbl>             <dbl>
	# 	1 nk_resting                        7.48               1.89
	# 2 nk_primed                         0.395              3.04
	# 3 t_gamma_delta                     0.375              3.34
	
	
	
})


# test_that("Check accuracy N52",{
# 
# 	N52_ARMET_T =
# 		filter(	readRDS("dev/N52.rds"),	ct == "T") %>%
# 		mutate(count = as.integer(count)) %>%
# 		ARMET_tc(
# 			.sample = sample,
# 			.transcript = symbol,
# 			.abundance = count
# 		) %>%
# 		ARMET_tc_continue(2) %>%
# 		ARMET_tc_continue(3)
# 
# 	expect_gt(
# 		N52_ARMET_T$proportions %>% filter(`Cell type category` == "immune_cell") %>% select(-.variable) %>%  summarise(.value %>% min),
# 		0.93
# 	)
# 
# 	expect_gt(
# 		N52_ARMET_T$proportions %>% filter(`Cell type category` == "t_cell") %>% select(-.variable) %>% summarise(.value %>% mean),
# 		0.79
# 	)
# 
# 	expect_lt(
# 		N52_ARMET_T$proportions %>% select(-.variable) %>% filter(!converged) %>% nrow,
# 		3
# 	)
# 
# })


# test_that("Simulated data and plot",{
# 
# res =
# 	readRDS("dev/noiseless_mix.rds") %>%
# 	gather(transcript, count, -sample, -covariate_2) %>%
# 	ARMET_tc(
# 		~ covariate_2,
# 		sample,
# 		transcript,
# 		count,
# 		iterations = 600,
# 		sampling_iterations = 400
# 	)  %>%
# 	ARMET_tc_continue(2) %>%
# 	ARMET_tc_continue(3)
# 
# expect_equal(
# 	c("b_cell" ,  "mast_cell" , "b_memory",   "neutrophil") %in%
# 	(res %>%
# 		test_differential_composition() %>%
# 		filter(significant) %>%
# 		pull(`Cell type category`) %>%
# 		unique
# 	) %>%
# 		all(),
# 	TRUE
# )
# 
# 
# # Test plotting
# res %>%
# 	test_differential_composition() %>%
# 	plot_polar()
# 
# })

# test_that("censoring",{
# 
# 	mix = readRDS("dev/mix_for_package_test.rds")
# 
# 	res =
# 		mix %>%
# 		mutate(`count mix` = as.integer(`count mix`)) %>%
# 		mutate(run = as.character(run)) %>%
# 		select(-level) %>%
# 
# 		ARMET_tc(
# 			 ~ censored(days, alive),
# 			run,
# 			symbol,
# 			`count mix`,
# 			iterations = 600,
# 			sampling_iterations = 400,
# 			prior_survival_time = mix %>% distinct(run, real_days) %>% pull(real_days)
# 
# 		) %>%
# 		ARMET_tc_continue(2) %>%
# 		ARMET_tc_continue(3)
# 
# 	expect_equal(
# 		c("immune_cell", "b_cell" ,     "b_memory" ,   "b_naive" ) %in%
# 			(res %>%
# 			 	test_differential_composition() %>%
# 			 	filter(significant) %>%
# 			 	pull(`Cell type category`) %>%
# 			 	unique
# 			) %>%
# 			all(),
# 		TRUE
# 	)
# 
# })