tanu/mosaic_2020
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

reformatting code to select needed df for analysis

Browse source
This commit is contained in:
Tanushree Tunstall 2020-11-20 11:43:03 +00:00
parent a6cbaab40a
commit b72c4df796
7 changed files with 243 additions and 102 deletions
Download patch file Download diff file Expand all files Collapse all files

41
flu_stats_unpaired_clinical.R
View file

@ -25,27 +25,28 @@ outfile_clinical_unpaired
# Unpaired stats for clinical data b/w groups: wilcoxon UNpaired analysis
# No correction required
########################################################################
str(clinical_df_ics)
numerical_cols = c("age"
#, "vl_pfu_ul_npa1"
, "vl_pfu_ul_npa1"
, "los"
, "onset2final"
, "onsfindeath"
, "onset_2_initial"
, "o2_sat_admis")
#, "onset_2_initial" # already bin
#, "o2_sat_admis"# already bin
)
metadata_cols = c("mosaic", "obesity")
clinical_df_numerical = clinical_df[, c(metadata_cols, numerical_cols)]
clinical_df_numerical = clinical_df_ics[, c(metadata_cols, numerical_cols)]
pivot_cols = metadata_cols
#pivot_cols = metadata_cols[!meta_data_cols%in%cols_to_omit];pivot_cols
expected_rows_clinical_lf = nrow(clinical_df_numerical) * (length(clinical_df_numerical) - length(pivot_cols)); expected_rows_clinical_lf
# lf data colnames
keycol <- "clinical_params"
valuecol <- "value"
gathercols <- c("age", "los", "onset2final", "onsfindeath", "onset_2_initial", "o2_sat_admis")
gathercols <- numerical_cols
clinical_lf = gather_(clinical_df_numerical, keycol, valuecol, gathercols)
@ -70,12 +71,15 @@ stats_un_clinical = compare_means(value~obesity
#, data = clinical_lf_comp
, paired = FALSE)
head(stats_un_clinical)
# rstatix
stat_df <- clinical_lf %>%
group_by(clinical_params) %>%
wilcox_test(value ~ obesity, paired = F) %>%
add_significance("p")
stat_df$p_format = round(stat_df$p, digits = 3)
stat_df
#----------------------------------------
# calculate n_obs for each clinical param: Overall
@ -101,31 +105,39 @@ n_all_gp = merge(n_all, n_gp
#----------------------------------------
# calculate n_obs for each clinical param: complete cases
#----------------------------------------
n_comp = data.frame(table(clinical_lf_comp$clinical_params))
n_comp = data.frame(table(clinical_lf$clinical_params))
colnames(n_comp) = c("clinical_params", "n_complete")
n_comp$clinical_params = as.character(n_comp$clinical_params)
n_comp
n_gp_comp_lf = data.frame(table(clinical_lf_comp$clinical_params, clinical_lf_comp$obesity)); n_gp_comp_lf
n_gp_comp_lf = data.frame(table(clinical_lf$clinical_params
, clinical_lf$obesity)); n_gp_comp_lf
n_gp_comp = spread(n_gp_comp_lf, "Var2", "Freq"); n_gp_comp
colnames(n_gp_comp)
colnames(n_gp_comp) = c("clinical_params"
, paste0("n_complete_gp", colnames(n_gp_comp)[2])
, paste0("n_complete_gp", colnames(n_gp_comp)[3]))
#---------
# merge 1
#---------
n_comp_gp = merge(n_comp, n_gp_comp
, by = intersect( names(n_comp), names(n_gp_comp))
, all = T)
n_comp_gp
#---------
# merge 2
#---------
merge_cols = intersect(names(n_all_gp), names(n_comp_gp)); merge_cols
n_df = merge(n_all_gp, n_comp_gp, by = merge_cols, all = T); n_df
#==================================
# Merge: merge stats + n_obs df
#===================================
#----------------------------------
# Merge 3: merge stats + n_obs df
#----------------------------------
merging_cols = intersect(names(stats_un_clinical), names(n_df)); merging_cols
if (all(n_df$clinical_params%in%stats_un_clinical$clinical_params)) {
cat("PASS: merging stats and n_obs on column/s:", merging_cols)
stats_un_clinical = merge(stats_un_clinical, n_df, by = merging_cols, all = T)
@ -188,6 +200,7 @@ if( length(my_col_order2) == ncol(stats_clinical_df) && (all(my_col_order2%in%co
quit()
}
# assign nice column names like replace "." with "_"
# same ordering as my_col_order2, just minor formatting
colnames(stats_clinical_df_f) = c("clinical_params"
, "method"
, "group1"
@ -208,4 +221,4 @@ colnames(stats_clinical_df_f)
# write output file
#******************
cat("UNpaired stats for clinical data for groups in:", outfile_clinical_unpaired)
#write.csv(stats_clinical_df_f, outfile_clinical_unpaired, row.names = FALSE)
write.csv(stats_clinical_df_f, outfile_clinical_unpaired, row.names = FALSE)