323 lines
12 KiB
R
323 lines
12 KiB
R
########################a###########################################################
|
|
# Input:
|
|
# Data
|
|
# mutable_df: merged_df3 containing the OR column to use as y-axis or any other relevant column
|
|
|
|
# x_axis_colname = "position"
|
|
# symbol_mut_colname = "mutant_type"
|
|
# symbol_wt_colname = "mutant_type"
|
|
# omit_snp_count = c(0, 1, 2...) can be used to filter positions with specified snp count
|
|
|
|
# my_logo_col = c("chemistry", "hydrophobicity", "clustalx", "taylor")
|
|
# --> if clustalx and taylor, set variable to black bg + white font
|
|
# --> if chemistry and hydrophobicity, then grey bg + black font
|
|
|
|
# ...other params
|
|
|
|
# Returns: Logo plot from combined data containing all SAVs per position.
|
|
# Helps to see the overview of SAV diversity
|
|
|
|
# TODO: SHINY
|
|
# select/drop down: omit_snp_count
|
|
# select/drop down: my_logo_col
|
|
# should include WT??
|
|
|
|
# Make it hover over position and then get the corresponding data table!
|
|
####################################################################################
|
|
|
|
#==================
|
|
# logo data: OR
|
|
#==================
|
|
# NOTE: my_logo_col
|
|
|
|
LogoPlotSnps <- function(plot_df
|
|
, x_axis_colname = "position"
|
|
, symbol_mut_colname = "mutant_type"
|
|
, symbol_wt_colname = "wild_type"
|
|
, omit_snp_count = c(0) # can be 1, 2, etc.
|
|
, my_logo_col = "chemistry"
|
|
, x_lab = "Position"
|
|
, y_lab = "SAV Count"
|
|
, x_ats = 6 # text size
|
|
, x_tangle = 90 # text angle
|
|
, y_ats = 10
|
|
, y_tangle = 0
|
|
, x_tts = 10 # title size
|
|
, y_tts = 10
|
|
, leg_pos = "none" # can be top, left, right and bottom or c(0.8, 0.9)
|
|
, leg_dir = "horizontal" #can be vertical or horizontal
|
|
, leg_ts = 10 # leg text size
|
|
, leg_tts = 8 # leg title size
|
|
, tpos0 = 0 # 0 is a magic number that does my sensible default
|
|
, tW0 = 1
|
|
, tH0 = 0.2
|
|
, debug=FALSE,
|
|
...
|
|
|
|
)
|
|
|
|
{
|
|
mutable_df=cbind(plot_df)
|
|
# handle funky omit_snp_count. DOES NOT WORK YET
|
|
if (class(omit_snp_count) != "numeric"){
|
|
omit_snp_count <- as.numeric(unlist(str_extract_all(omit_snp_count, regex("[0-9]+"))))
|
|
}
|
|
############################################
|
|
# Data processing for logo plot for SAVS
|
|
############################################
|
|
|
|
# Generate "ligand distance" colour map
|
|
# mutable_df = generate_distance_colour_map(mutable_df, debug=TRUE)
|
|
# unique_colour_map = unique(mutable_df[,c("position","ligD_colours")])
|
|
# unique_colour_map = unique_colour_map[order(unique_colour_map$position), ]
|
|
# rownames(unique_colour_map) = unique_colour_map$position
|
|
# unique_colour_map2 = unique_colour_map
|
|
# unique_colour_map2$position=as.factor(unique_colour_map2$position)
|
|
# unique_colour_map2$ligD_colours = as.factor(unique_colour_map2$ligD_colours)
|
|
#
|
|
|
|
setDT(mutable_df)[, mut_pos_occurrence := .N, by = .(eval(parse(text=x_axis_colname)))]
|
|
if (debug) {
|
|
table(mutable_df[[x_axis_colname]])
|
|
table(mutable_df$mut_pos_occurrence)
|
|
}
|
|
max_mut = max(table(mutable_df[[x_axis_colname]]))
|
|
|
|
# Subset Data as specified by user
|
|
cat("\nDisplaying SAV position frequency:\n")
|
|
print(table(mutable_df$mut_pos_occurrence))
|
|
|
|
if ( (length(omit_snp_count) ==1) && (omit_snp_count == 0) ){
|
|
my_data_snp = mutable_df
|
|
u = unique(my_data_snp[[x_axis_colname]])
|
|
max_mult_mut = max(table(my_data_snp[[x_axis_colname]]))
|
|
if (debug) {
|
|
cat("\nNo filtering requested:"
|
|
, "\nTotal no. of SAVs:", sum(table(mutable_df$mut_pos_occurrence))
|
|
, "\nTotal no. of SAVs omitted:", sum(table(mutable_df$mut_pos_occurrence)[omit_snp_count])
|
|
, "\nDim of data:", dim(my_data_snp)
|
|
, "\nNo. of positions:", length(u)
|
|
, "\nMax no. of muts at any position:", max_mult_mut)
|
|
}
|
|
} else {
|
|
|
|
my_data_snp = subset(mutable_df, !(mut_pos_occurrence%in%omit_snp_count) )
|
|
|
|
exp_nrows = sum(table(mutable_df$mut_pos_occurrence)) - sum(table(mutable_df$mut_pos_occurrence)[omit_snp_count])
|
|
got_rows = sum(table(my_data_snp$mut_pos_occurrence))
|
|
u = unique(my_data_snp[[x_axis_colname]])
|
|
max_mult_mut = max(table(my_data_snp[[x_axis_colname]]))
|
|
if (debug) {
|
|
if (got_rows == exp_nrows) {
|
|
cat("\nPass: Position with the stated SAV frequency filtered:", omit_snp_count
|
|
, "\nTotal no. of SAVs:", sum(table(mutable_df$mut_pos_occurrence))
|
|
, "\nTotal no. of SAVs omitted:", sum(table(mutable_df$mut_pos_occurrence)[omit_snp_count])
|
|
, "\nDim of subsetted data:", dim(my_data_snp)
|
|
, "\nNo. of positions:", length(u)
|
|
, "\nMax no. of muts at any position:", max_mult_mut)
|
|
} else {
|
|
|
|
cat("\nFAIL:Position with the stated SAV frequency COULD NOT be filtered..."
|
|
, "\nExpected:",exp_nrows
|
|
, "\nGot:", got_rows )
|
|
}
|
|
}
|
|
}
|
|
|
|
#--------------------------------------
|
|
# matrix for mutant type
|
|
# frequency of mutant type by position
|
|
#---------------------------------------
|
|
table(my_data_snp[[symbol_mut_colname]], my_data_snp[[x_axis_colname]])
|
|
tab_mt = table(my_data_snp[[symbol_mut_colname]], my_data_snp[[x_axis_colname]])
|
|
class(tab_mt)
|
|
|
|
# unclass to convert to matrix
|
|
tab_mt = unclass(tab_mt)
|
|
|
|
if (is.matrix(tab_mt)){
|
|
if (debug) {
|
|
cat("\nPASS: Mutant matrix successfully created..."
|
|
#, "\nRownames of mutant matrix:", rownames(tab_mt)
|
|
#, "\nColnames of mutant matrix:", colnames(tab_mt)
|
|
)
|
|
}
|
|
} else{
|
|
tab_mt = as.matrix(tab_mt, rownames = T)
|
|
if (is.matrix(tab_mt)){
|
|
if (debug) {
|
|
cat("\nCreating mutant matrix..."
|
|
#, "\nRowna mes of mutant matrix:", rownames(tab_mt)
|
|
#, "\nColnames of mutant matrix:", colnames(tab_mt)
|
|
)
|
|
}
|
|
}
|
|
}
|
|
|
|
#-------------------------------------
|
|
# matrix for wild type
|
|
# frequency of wild type by position
|
|
#-------------------------------------
|
|
tab_wt = table(my_data_snp[[symbol_wt_colname]], my_data_snp[[x_axis_colname]]); tab_wt
|
|
tab_wt = unclass(tab_wt)
|
|
|
|
# Important: remove wt duplicates
|
|
#wt = my_data_snp[, c("position", "wild_type")]
|
|
wt = my_data_snp %>%
|
|
select(x_axis_colname, symbol_wt_colname)
|
|
|
|
wt = wt[!duplicated(wt),]
|
|
wt
|
|
|
|
tab_wt = table(wt[[symbol_wt_colname]], wt[[x_axis_colname]]); tab_wt # should all be 1
|
|
if (debug) {
|
|
if ( identical(colnames(tab_mt), colnames(tab_wt) ) && identical(ncol(tab_mt), ncol(tab_wt)) ){
|
|
|
|
cat("\nPASS: Wild type matrix successfully created"
|
|
, "\nDim of wt matrix:", dim(tab_wt)
|
|
, "\nDim of mutant matrix:", dim(tab_mt)
|
|
, "\n"
|
|
#, "\nRownames of mutant matrix:", rownames(tab_wt)
|
|
#, "\nColnames of mutant matrix:", colnames(tab_wt)
|
|
)
|
|
}
|
|
}
|
|
|
|
######################################
|
|
# Generating plots for muts and wt
|
|
#####################################
|
|
LogoPlotL <- list()
|
|
|
|
if (my_logo_col %in% c('clustalx','taylor')) {
|
|
cat("\nSelected colour scheme:", my_logo_col
|
|
, "\nUsing black theme\n")
|
|
|
|
theme_bgc = "black"
|
|
xfont_bgc = "white"
|
|
yfont_bgc = "white"
|
|
xtt_col = "white"
|
|
ytt_col = "white"
|
|
}
|
|
|
|
if (my_logo_col %in% c('chemistry', 'hydrophobicity')) {
|
|
cat('\nSelected colour scheme:', my_logo_col
|
|
, "\nUsing grey theme")
|
|
|
|
theme_bgc = "white"
|
|
xfont_bgc = "black"
|
|
yfont_bgc = "black"
|
|
xtt_col = "black"
|
|
ytt_col = "black"
|
|
}
|
|
position_mt = as.numeric(colnames(tab_mt))
|
|
position_wt = as.numeric(colnames(tab_wt))
|
|
#####################################
|
|
# Generating logo plots for SAVs
|
|
#####################################
|
|
#-------------------
|
|
# Mutant logo plot
|
|
#-------------------
|
|
logo_top = ggplot() +
|
|
geom_logo(tab_mt
|
|
, method = 'custom'
|
|
, col_scheme = my_logo_col
|
|
, seq_type = 'aa') +
|
|
theme_nothing() +
|
|
ylab(y_lab) +
|
|
theme(text=element_text(family="FreeSans")
|
|
, legend.position = leg_pos
|
|
, legend.direction = leg_dir
|
|
, legend.title = element_text(size = leg_tts
|
|
, colour = ytt_col)
|
|
, legend.text = element_text(size = leg_ts)
|
|
|
|
, axis.text.x = element_text(size = x_ats
|
|
, angle = x_tangle
|
|
#, hjust = 1
|
|
#, vjust = 0.4
|
|
, colour = xfont_bgc
|
|
#, margin = margin(t = 0.1)
|
|
)
|
|
, axis.text.y = element_blank()
|
|
# , axis.text.y = element_text(size = y_ats
|
|
# , angle = y_tangle
|
|
# , hjust = 1
|
|
# , vjust = -1.0
|
|
# , colour = yfont_bgc)
|
|
# , axis.title.x = element_text(size = x_tts
|
|
# , colour = xtt_col)
|
|
, axis.title.x = element_blank()
|
|
, axis.title.y = element_text(size = y_tts
|
|
, angle = 90
|
|
, colour = ytt_col
|
|
, margin = margin(t = 0, r = 0, b = 20, l = 0)
|
|
#, hjust = -2
|
|
)
|
|
|
|
, plot.background = element_rect(fill = theme_bgc, colour=NA)
|
|
) +
|
|
scale_x_discrete("Position"
|
|
, labels = position_mt
|
|
, limits = factor(1:length(position_mt))
|
|
)
|
|
|
|
logo_bottom = ggplot() +
|
|
geom_logo(tab_wt
|
|
, method = 'custom'
|
|
, col_scheme = my_logo_col
|
|
, seq_type = 'aa') +
|
|
theme_nothing() +
|
|
scale_x_discrete("Position"
|
|
, labels = x_axis_colname
|
|
, limits = factor(1:length(x_axis_colname))) +
|
|
theme(text = element_text(family="FreeSans")
|
|
, legend.position = "none"
|
|
#, axis.text.x = element_blank()
|
|
#, axis.text.y = element_blank()
|
|
#, axis.text.y = element_text()
|
|
, axis.title.x = element_blank()
|
|
#, axis.title.y = element_blank()
|
|
, axis.title.y = element_text(size = y_tts
|
|
, angle = 90
|
|
, colour = ytt_col
|
|
, margin = margin(t = 0, r = 0, b = 20, l = 0))
|
|
, plot.background = element_rect(fill = theme_bgc, colour=NA)
|
|
) +
|
|
labs(x=NULL, y="WT")
|
|
|
|
anno_bar = position_annotation(plot_df,
|
|
bg = theme_bgc,
|
|
# active_aa_pos = active_aa_pos,
|
|
# aa_pos_drug = aa_pos_drug,
|
|
# aa_pos_lig1 = aa_pos_lig1,
|
|
# aa_pos_lig2 = aa_pos_lig2,
|
|
# aa_pos_lig3 = aa_pos_lig3,
|
|
...
|
|
)
|
|
|
|
#aligned=align_plots(logo_top, logo_bottom, anno_bar, align='vh', axis='lr')
|
|
cowplot::plot_grid(
|
|
logo_top, logo_bottom, anno_bar,
|
|
#aligned[[1]], aligned[[2]], aligned[[3]],
|
|
ncol=1,
|
|
align = "v",
|
|
rel_heights = c(7, 1,1)
|
|
)
|
|
|
|
# cowplot::plot_grid(
|
|
# logo_top,
|
|
# #NULL,
|
|
# logo_bottom,
|
|
# #NULL,
|
|
# anno_bar,
|
|
# ncol=1,
|
|
# align = "v",
|
|
# rel_heights = c(7, 1,1)
|
|
# )
|
|
# top logo, bottom logo, heat bar, NULL, position annotation
|
|
#------------------
|
|
# Wild logo plot
|
|
#------------------
|
|
}
|
|
|
|
#LogoPlotSnps(small_df3)
|