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

refactoring logo plots to add flame bar

Browse source
This commit is contained in:
Tanushree Tunstall 2022-08-06 18:47:54 +01:00
parent f0a9eb4eec
commit c968089cd2
4 changed files with 194 additions and 196 deletions
Download patch file Download diff file Expand all files Collapse all files

83
scripts/functions/generate_distance_colour_map.R
View file

@ -1,64 +1,29 @@
# takes a dataframe and returns the same dataframe with two extra columns for colours and position # takes a dataframe and returns the same dataframe with two extra columns for colours and position
generate_distance_colour_map = function(plotdf, generate_distance_colour_map = function(plot_df,
xvar_colname = "position", xvar_colname = "position",
yvar_colname = 'duet_scaled',
lig_dist_colname = "ligand_distance", lig_dist_colname = "ligand_distance",
lig_dist_colours = c("green", "yellow", "orange", "red"), lig_dist_colours = c("green", "yellow", "orange", "red"),
#tpos0 = 0, debug = TRUE
#tpos1 = 0,
#tpos2 = 0,
#tpos3 = 0,
debug = FALSE
) )
{ {
#-------------------
# x and y axis
# range, scale, etc
#-------------------
my_xlim = length(unique(plotdf[[yvar_colname]])); my_xlim
ymin = min(plotdf[[yvar_colname]]); ymin
ymax = max(plotdf[[yvar_colname]]); ymax
#if (tpos0 == 0){
# tpos0 = ymin-0.5
#}
#if (tpos1 == 0){
# tpos1 = ymin-0.65
#}
#if (tpos2 == 0){
# tpos2 = ymin-0.75
#}
#if (tpos3 == 0){
# tpos3 = ymin-0.85
#}
if (debug) { if (debug) {
cat("\nAnnotating x-axis ~", lig_dist_colname, "requested...") cat("\nAnnotating x-axis ~", lig_dist_colname, "requested...")
} }
#-------------------------------------
# round column values: to colour by
#--------------------------------------
#plotdf = plotdf[order(plotdf[[lig_dist_colname]]),]
plotdf['lig_distR'] = round(plotdf[[lig_dist_colname]], digits = 0)
#head(plotdf['lig_distR'])
#------------------------------------- plot_df['lig_distR'] = round(plot_df[[lig_dist_colname]], digits = 0)
# ligand distance range, min, max, etc
#-------------------------------------- lig_min = min(round(plot_df[[lig_dist_colname]]), na.rm = T); lig_min
lig_min = min(round(plotdf[[lig_dist_colname]]), na.rm = T); lig_min lig_max = max(round(plot_df[[lig_dist_colname]]), na.rm = T); lig_max
lig_max = max(round(plotdf[[lig_dist_colname]]), na.rm = T); lig_max lig_mean = round(mean(round(plot_df[[lig_dist_colname]]), na.rm = T)); lig_mean
lig_mean = round(mean(round(plotdf[[lig_dist_colname]]), na.rm = T)); lig_mean
#-------------------------------------
# Create mapping colour key # Create mapping colour key
#-------------------------------------- n_colours = length(sort(unique(round(plot_df[[lig_dist_colname]], digits = 0)))); n_colours
# sorting removes NA, so that n_colours == length(ligD_valsR)
n_colours = length(sort(unique(round(plotdf[[lig_dist_colname]], digits = 0)))); n_colours
lig_cols = colorRampPalette(lig_dist_colours)(n_colours); lig_cols lig_cols = colorRampPalette(lig_dist_colours)(n_colours); lig_cols
ligD_valsR = sort(unique(round(plotdf[[lig_dist_colname]], digits = 0))); ligD_valsR ligD_valsR = sort(unique(round(plot_df[[lig_dist_colname]], digits = 0))); ligD_valsR
length(ligD_valsR)
if (debug) { if (debug) {
length(ligD_valsR)
if (n_colours == length(ligD_valsR)) { if (n_colours == length(ligD_valsR)) {
cat("\nStarting: mapping b/w" cat("\nStarting: mapping b/w"
, lig_dist_colname , lig_dist_colname
@ -73,23 +38,23 @@ generate_distance_colour_map = function(plotdf,
ligDcolKey <- data.frame(ligD_colours = lig_cols ligDcolKey <- data.frame(ligD_colours = lig_cols
, lig_distR = ligD_valsR); ligDcolKey , lig_distR = ligD_valsR); ligDcolKey
names(ligDcolKey)
if (debug) { if (debug) {
names(ligDcolKey)
cat("\nSuccessful: Mapping b/w", lig_dist_colname, "and colours") cat("\nSuccessful: Mapping b/w", lig_dist_colname, "and colours")
} }
#------------------------------------- #-------------------------------------
# merge colour key with plotdf # merge colour key with plot_df
#-------------------------------------- #--------------------------------------
plotdf = merge(plotdf, ligDcolKey, by = 'lig_distR') plot_df = merge(plot_df, ligDcolKey, by = 'lig_distR')
plotdf_check = as.data.frame(cbind(position = plotdf[[xvar_colname]] # plot_df_check = as.data.frame(cbind(position = plot_df[[xvar_colname]]
, ligD = plotdf[[lig_dist_colname]] # , ligD = plot_df[[lig_dist_colname]]
, ligDR = plotdf$lig_distR # , ligDR = plot_df$lig_distR
, ligD_cols = plotdf$ligD_colours)) # , ligD_cols = plot_df$ligD_colours))
return(plotdf) return(plot_df)
} }
generate_distance_legend = function(plotdf, generate_distance_legend = function(plot_df,
yvar_colname, yvar_colname,
xvar_colname = 'position', xvar_colname = 'position',
lig_dist_colname = "ligand_distance", lig_dist_colname = "ligand_distance",
@ -97,15 +62,15 @@ generate_distance_legend = function(plotdf,
) )
{ {
# build legend for ligand distance "heat bar" # build legend for ligand distance "heat bar"
lig_min = min(round(plotdf[[lig_dist_colname]]), na.rm = T); lig_min lig_min = min(round(plot_df[[lig_dist_colname]]), na.rm = T); lig_min
lig_max = max(round(plotdf[[lig_dist_colname]]), na.rm = T); lig_max lig_max = max(round(plot_df[[lig_dist_colname]]), na.rm = T); lig_max
lig_mean = round(mean(round(plotdf[[lig_dist_colname]]), na.rm = T)); lig_mean lig_mean = round(mean(round(plot_df[[lig_dist_colname]]), na.rm = T)); lig_mean
labels = seq(lig_min, lig_max, len = 5); labels labels = seq(lig_min, lig_max, len = 5); labels
labelsD = round(labels, digits = 0); labelsD labelsD = round(labels, digits = 0); labelsD
get_legend(ggplot(plotdf, aes_string(x = sprintf("factor(%s)", xvar_colname) get_legend(
, y = yvar_colname)) + ggplot(plot_df, aes_string(x = sprintf("factor(%s)", xvar_colname), y=0)) +
geom_tile(aes(fill = .data[[lig_dist_colname]]) geom_tile(aes(fill = .data[[lig_dist_colname]])
, colour = "white") + , colour = "white") +

33
scripts/functions/lf_bp2.R
View file

@ -54,6 +54,7 @@ lf_bp2 <- function(lf_df = lf_duet
}else{ }else{
bp_width = bp_width bp_width = bp_width
} }
my_comparisonsL <- list( stat_grp_comp )
ggplot(lf_df, aes_string(x = x_grp, y = y_var)) + ggplot(lf_df, aes_string(x = x_grp, y = y_var)) +
@ -93,39 +94,11 @@ lf_bp2 <- function(lf_df = lf_duet
, x = "" , x = ""
, y = "") + , y = "") +
if (add_stats){
my_comparisonsL <- list( stat_grp_comp )
stat_compare_means(comparisons = my_comparisonsL stat_compare_means(comparisons = my_comparisonsL
, method = stat_method , method = stat_method
, paired = my_paired , paired = my_paired
, label = stat_label[2]) , label = stat_label[2]) +
if (make_boxplot){
geom_boxplot(fill = "white"
, outlier.colour = NA
#, position = position_dodge(width = 0.9)
, width = bp_width) +
geom_quasirandom(#priority = "density"
#, shape = 21
size = dot_size
, alpha = dot_transparency
, show.legend = FALSE
, cex = 0.8
, aes(
colour = factor(
eval(
parse(
text = colour_categ
)
)
)
)
)
} else {
geom_quasirandom( geom_quasirandom(
size = dot_size size = dot_size
, alpha = dot_transparency , alpha = dot_transparency
@ -143,7 +116,5 @@ lf_bp2 <- function(lf_df = lf_duet
) )
) )
} }
}
}
#lf_bp2(lf_consurf) #lf_bp2(lf_consurf)

96
scripts/functions/logoP_or.R
View file

@ -46,7 +46,7 @@ LogoPlotCustomH <- function(plot_df
, rm_empty_y = F , rm_empty_y = F
, y_axis_log = F , y_axis_log = F
, log_value = log10 , log_value = log10
, y_axis_increment = 5 , y_axis_increment = 50
, x_lab = "Position" , x_lab = "Position"
, y_lab = "Odds Ratio" , y_lab = "Odds Ratio"
, x_ats = 12 # text size , x_ats = 12 # text size
@ -68,16 +68,24 @@ LogoPlotCustomH <- function(plot_df
################################# #################################
# Data processing for logo plot # Data processing for logo plot
################################# #################################
plot_df = generate_distance_colour_map(plot_df, yvar_colname = y_axis_colname, debug=TRUE) plot_df = generate_distance_colour_map(plot_df, debug=TRUE)
unique_colour_map = unique(plot_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)
if (rm_empty_y){ if (rm_empty_y){
plot_df = plot_df[!is.na(plot_df[y_axis_colname]),] plot_df = plot_df[!is.na(plot_df[y_axis_colname]),]
cat("\nRemoving empty positions...\n") cat("\nRemoving empty positions...\n")
}else{
plot_df = plot_df
} }
y_max = max(plot_df['or_mychisq'], na.rm = T) y_max = max(plot_df[[y_axis_colname]], na.rm = T)
cat("\nRemoving y scale incremenet:", y_axis_increment) cat("\nRemoving y scale incremenet:", y_axis_increment)
y_lim = round_any(y_max, y_axis_increment, f = ceiling) y_lim = round_any(y_max, y_axis_increment, f = ceiling)
@ -155,32 +163,44 @@ LogoPlotCustomH <- function(plot_df
y_lab = paste("Log", y_lab) y_lab = paste("Log", y_lab)
} }
} }
ggseqlogo(logo_dfP_wf plot_grid(
ggplot() +
geom_logo(logo_dfP_wf
, method = "custom" , method = "custom"
, col_scheme = my_logo_col , col_scheme = my_logo_col
, seq_type = "aa") + , seq_type = "aa") +
#ylab("my custom height") + #ylab("my custom height") +
theme(axis.text.x = element_text(size = x_ats theme( axis.ticks.x = element_blank()
, angle = x_tangle , axis.title.x = element_blank()
, hjust = 1 , axis.text.x = element_blank() # turn this off and the below on if you want to visually
, vjust = 0.4 # verify positions.
, colour = xfont_bgc) # , axis.text.x = element_text(size = x_ats
# , angle = x_tangle
# , hjust = 1
# , vjust = 0.4
# , colour = xfont_bgc
# , margin = margin(b=0)
# )
, axis.text.y = element_text(size = y_ats , axis.text.y = element_text(size = y_ats
, angle = y_tangle , angle = y_tangle
, hjust = 1 , hjust = 1
, vjust = 0 , vjust = 0
, colour = yfont_bgc) , colour = yfont_bgc)
, axis.title.x = element_text(size = x_tts
, colour = xtt_col)
, axis.title.y = element_text(size = y_tts , axis.title.y = element_text(size = y_tts
, colour = ytt_col) , colour = ytt_col)
, legend.title = element_text(size = leg_tts , legend.title = element_text(size = leg_tts
, colour = ytt_col) , colour = ytt_col)
, legend.text = element_text(size = leg_ts) #, legend.text = element_text(size = leg_ts)
, legend.text = element_blank()
, legend.position = leg_pos , legend.position = leg_pos
, legend.direction = leg_dir , legend.direction = leg_dir
, plot.background = element_rect(fill = theme_bgc))+ , plot.background = element_rect(fill = theme_bgc)
, plot.margin = margin(b=0)
, panel.grid=element_blank()
, panel.background = element_rect(fill = theme_bgc)
)+
scale_x_discrete(x_lab scale_x_discrete(x_lab
#, breaks #, breaks
@ -191,14 +211,42 @@ LogoPlotCustomH <- function(plot_df
, breaks = seq(0, (y_lim), by = y_axis_increment) , breaks = seq(0, (y_lim), by = y_axis_increment)
#, labels = seq(0, (y_lim), by = y_axis_increment) #, labels = seq(0, (y_lim), by = y_axis_increment)
, limits = c(0, y_lim)) + , limits = c(0, y_lim)) +
ylab(y_lab) + labs(y=y_lab),
geom_tile(aes(plot_df$position, tpos0 # heat-mapped distance tiles along the bot ggplot(data=unique_colour_map2, aes(
, width = tW0 x=factor(position), 0 # heat-mapped distance tiles along the bot
, height = tH0) , fill = position
, fill = plot_df$ligD_colours , colour = position
, colour = plot_df$ligD_colours , linetype = "blank"
, linetype = "blank") )
#LogoPlot = p0 + ylab(y_lab) ) +
#return(LogoPlot) geom_tile() +
theme(axis.text.x = element_text(size = x_ats
, angle = x_tangle
, hjust = 1
, vjust = 0.4
, colour = xfont_bgc)
, axis.text.y = element_blank()
, axis.ticks.y = element_blank()
, axis.title.x = element_text(size = x_tts
, colour = xtt_col)
, axis.title.y = element_text(size = y_tts
, colour = ytt_col)
, legend.title = element_text(size = leg_tts
, colour = ytt_col)
, legend.text = element_text(size = leg_ts)
, legend.position = leg_pos
, legend.direction = leg_dir
, plot.background = element_rect(fill = theme_bgc)
, plot.margin = margin(t=0)
, panel.grid=element_blank()
, panel.background = element_rect(fill = theme_bgc)
) +
scale_x_discrete(x_lab, labels=factor(unique_colour_map$position)) +
scale_color_manual(values=unique_colour_map$ligD_colours) +
scale_fill_manual(values=unique_colour_map$ligD_colours) +
labs(y = NULL),
ncol=1, align='v', rel_heights = c(8/10,1/10)
)
} }
#LogoPlotCustomH(merged_df3)

80
scripts/functions/logoP_snp.R
View file

@ -33,7 +33,7 @@
LogoPlotSnps <- function(plot_df LogoPlotSnps <- function(plot_df
, x_axis_colname = "position" , x_axis_colname = "position"
, symbol_mut_colname = "mutant_type" , symbol_mut_colname = "mutant_type"
, symbol_wt_colname = "mutant_type" , symbol_wt_colname = "wild_type"
, omit_snp_count = c(0) # can be 1, 2, etc. , omit_snp_count = c(0) # can be 1, 2, etc.
, my_logo_col = "chemistry" , my_logo_col = "chemistry"
, x_lab = "Position" , x_lab = "Position"
@ -66,6 +66,13 @@ LogoPlotSnps <- function(plot_df
# Generate "ligand distance" colour map # Generate "ligand distance" colour map
plot_df = generate_distance_colour_map(plot_df, debug=TRUE) plot_df = generate_distance_colour_map(plot_df, debug=TRUE)
unique_colour_map = unique(plot_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(plot_df)[, mut_pos_occurrence := .N, by = .(eval(parse(text=x_axis_colname)))] setDT(plot_df)[, mut_pos_occurrence := .N, by = .(eval(parse(text=x_axis_colname)))]
if (debug) { if (debug) {
@ -213,10 +220,6 @@ LogoPlotSnps <- function(plot_df
, col_scheme = my_logo_col , col_scheme = my_logo_col
, seq_type = 'aa') + , seq_type = 'aa') +
theme(text=element_text(family="FreeSans"))+
theme(axis.text.x = element_blank()) +
theme_logo()+
scale_x_continuous(breaks = 1:ncol(tab_mt) scale_x_continuous(breaks = 1:ncol(tab_mt)
, expand = c(0.01,0) , expand = c(0.01,0)
, labels = colnames(tab_mt))+ , labels = colnames(tab_mt))+
@ -224,16 +227,9 @@ LogoPlotSnps <- function(plot_df
scale_y_continuous(breaks = 0:(max_mult_mut-1) scale_y_continuous(breaks = 0:(max_mult_mut-1)
, labels = c(1:max_mult_mut) , labels = c(1:max_mult_mut)
, limits = c(0, max_mult_mut)) + , limits = c(0, max_mult_mut)) +
# FIXME: currently broken, possibly due to ggseqlogo() not working in the
# way standard ggplot2() does
geom_tile(aes(plot_df$position, tpos0 # heat-mapped distance tiles along the bottom.
, width = tW0
, height = tH0)
, fill = plot_df$ligD_colours
, colour = plot_df$ligD_colours
, linetype = "blank") +
ylab(y_lab) + ylab(y_lab) +
theme(legend.position = leg_pos theme(text=element_text(family="FreeSans")
, legend.position = leg_pos
, legend.direction = leg_dir , legend.direction = leg_dir
, legend.title = element_text(size = leg_tts , legend.title = element_text(size = leg_tts
, colour = ytt_col) , colour = ytt_col)
@ -260,40 +256,58 @@ LogoPlotSnps <- function(plot_df
, method = 'custom' , method = 'custom'
, col_scheme = my_logo_col , col_scheme = my_logo_col
, seq_type = 'aa') + , seq_type = 'aa') +
theme(text = element_text(family="FreeSans"))+
theme(axis.text.x = element_blank()
, axis.text.y = element_blank()) +
theme_logo()+
scale_x_continuous(breaks = 1:ncol(tab_wt) scale_x_continuous(breaks = 1:ncol(tab_wt)
, expand = c(0.01,0) , expand = c(0.01,0)
, labels = colnames(tab_wt))+ , labels = colnames(tab_wt))+
theme(text = element_text(family="FreeSans")
xlab(x_lab) + , legend.position = "none"
, axis.text.x = element_blank()
theme(legend.position = "none" , axis.text.y = element_blank()
, legend.direction = leg_dir , axis.title.x = element_blank()
#, legend.title = element_blank() , axis.title.y = element_blank()
, legend.title = element_text(size = y_tts , plot.background = element_rect(fill = theme_bgc)
, colour = ytt_col) ) +
, legend.text = element_text(size = leg_ts) labs(x=NULL, y=NULL)
, axis.text.x = element_text(size = x_ats ,
ggplot(data=unique_colour_map2, aes(
x=factor(position), 0 # heat-mapped distance tiles along the bot
, fill = position
, colour = position
, linetype = "blank"
)
) +
geom_tile() +
theme(axis.text.x = element_text(size = x_ats
, angle = x_tangle , angle = x_tangle
, hjust = 1 , hjust = 1
, vjust = 0.4 , vjust = 0.4
, colour = xfont_bgc) , colour = xfont_bgc)
, axis.text.y = element_blank() , axis.text.y = element_blank()
, axis.ticks.y = element_blank()
, axis.title.x = element_text(size = x_tts , axis.title.x = element_text(size = x_tts
, colour = xtt_col) , colour = xtt_col)
, axis.title.y = element_text(size = y_tts , axis.title.y = element_text(size = y_tts
, colour = ytt_col) , colour = ytt_col)
, legend.title = element_text(size = leg_tts
, colour = ytt_col)
, legend.text = element_text(size = leg_ts)
, legend.position = leg_pos
, legend.direction = leg_dir
, plot.background = element_rect(fill = theme_bgc) , plot.background = element_rect(fill = theme_bgc)
) , plot.margin = margin(t=0)
, nrow = 2 , panel.grid=element_blank()
, panel.background = element_rect(fill = theme_bgc)
) +
scale_x_discrete(x_lab, labels=factor(unique_colour_map$position)) +
scale_color_manual(values=unique_colour_map$ligD_colours) +
scale_fill_manual(values=unique_colour_map$ligD_colours) +
labs(y = NULL)
, nrow = 3
, align = "v" , align = "v"
, rel_heights = c(3/4, 1/4)) , rel_heights = c(7/10, 2/7, 1/7))
#------------------ #------------------
# Wild logo plot # Wild logo plot
#------------------ #------------------
} }
#LogoPlotSnps(mutable_df3)