playing with MSA plots to allow filtering of positions, arghhh
This commit is contained in:
parent
08bd8a2ee5
commit
00094f036a
6 changed files with 209 additions and 46 deletions
|
|
@ -20,33 +20,116 @@
|
|||
|
||||
LogoPlotMSA <- function(msaSeq_mut
|
||||
, msaSeq_wt
|
||||
, plot_positions
|
||||
, msa_method = 'bits' # or probability
|
||||
, my_logo_col = "chemistry"
|
||||
, x_lab = "Wild-type position"
|
||||
, y_lab = "Count"
|
||||
, x_ats = 13 # text size
|
||||
, x_tangle = 90 # text angle
|
||||
, y_ats = 13
|
||||
, y_tangle = 0
|
||||
, x_tts = 13 # title size
|
||||
, y_tts = 13
|
||||
, leg_pos = "top" # can be top, left, right and bottom or c(0.8, 0.9)
|
||||
, leg_dir = "horizontal" #can be vertical or horizontal
|
||||
, leg_ts = 16 # leg text size
|
||||
, leg_tts = 16 # leg title size
|
||||
, my_logo_col = "chemistry"
|
||||
, x_lab = "Wild-type position"
|
||||
, y_lab = ""
|
||||
, x_ats = 13 # text size
|
||||
, x_tangle = 90 # text angle
|
||||
, y_ats = 13
|
||||
, y_tangle = 0
|
||||
, x_tts = 13 # title size
|
||||
, y_tts = 13
|
||||
, leg_pos = "top" # can be top, left, right and bottom or c(0.8, 0.9)
|
||||
, leg_dir = "horizontal" #can be vertical or horizontal
|
||||
, leg_ts = 16 # leg text size
|
||||
, leg_tts = 16 # leg title size
|
||||
)
|
||||
|
||||
{
|
||||
############################################
|
||||
# Data processing for logo plot for nsSNPS
|
||||
############################################
|
||||
|
||||
############################################
|
||||
# Data processing for logo plot for nsSNPS
|
||||
###########################################
|
||||
cat("\nLength of MSA", length(msaSeq_mut)
|
||||
, "\nlength of WT seq:", length(msaSeq_wt))
|
||||
|
||||
if(missing(plot_positions)){
|
||||
#if(is.null(plot_positions)){
|
||||
cat("\nPlotting entire MSA")
|
||||
msa_seq_plot = msaSeq_mut
|
||||
wt_seq_plot = msaSeq_wt
|
||||
|
||||
} else {
|
||||
cat("\nUser specified plotting positions for MSA:"
|
||||
, "These are:", plot_positions)
|
||||
|
||||
cat("\nLength of MSA", nrow(msaSeq_mut)
|
||||
, "\nlength of WT seq:", nrow(msaSeq_wt))
|
||||
#-----------
|
||||
# MSA: mut
|
||||
#-----------
|
||||
cat("\nGenerating MSA: filtered positions")
|
||||
msa_interim = sapply(msaSeq_mut, function(x) unlist(strsplit(x,"")))
|
||||
|
||||
######################################
|
||||
# Generating plots for muts and wt
|
||||
#####################################
|
||||
if (any(is.na(msa_interim[plot_positions]))){
|
||||
cat("Plot_positions selected:", length(plot_positions))
|
||||
i_ofr = plot_positions[is.na(msa_interim[plot_positions])]
|
||||
cat("\nIndex out of range: 1 or more"
|
||||
, "\nThese are:", i_ofr
|
||||
, "\nOmitting these and proceeding...")
|
||||
i_extract = na.omit(msa_interim[plot_positions])
|
||||
cat("\nFinal positions being plottted:", length(i_extract)
|
||||
, "\nNo. of positions dropped from request:", length(i_ofr))
|
||||
|
||||
}else{
|
||||
cat("\nAll positions within range"
|
||||
, "\nProceeing with generating requested position MSA seqs...")
|
||||
i_extract = plot_positions
|
||||
}
|
||||
|
||||
matP1 = msa_interim[i_extract, 1:ncol(msa_interim)]
|
||||
|
||||
dfP1 = data.frame(t(matP1))
|
||||
names(dfP1) = i_extract
|
||||
cols_to_paste = names(dfP1)
|
||||
dfP1['chosen_seq'] = apply( dfP1[ , cols_to_paste]
|
||||
, 1
|
||||
, paste, sep = ''
|
||||
, collapse = "")
|
||||
|
||||
msa_seq_plot = dfP1$chosen_seq
|
||||
|
||||
#-----------
|
||||
# WT: fasta
|
||||
#-----------
|
||||
cat("\nGenerating WT fasta: filtered positions")
|
||||
|
||||
wt_interim = sapply(msaSeq_wt, function(x) unlist(strsplit(x,"")))
|
||||
|
||||
if (any(is.na(wt_interim[plot_positions]))){
|
||||
cat("Plot_positions selected:", length(plot_positions))
|
||||
i2_ofr = plot_positions[is.na(wt_interim[plot_positions])]
|
||||
cat("\nIndex out of range: 1 or more"
|
||||
, "\nThese are:", i2_ofr
|
||||
, "\nOmitting these and proceeding...")
|
||||
i2_extract = na.omit(wt_interim[plot_positions])
|
||||
cat("\nFinal positions being plottted:", length(i2_extract)
|
||||
, "\nNo. of positions dropped from request:", length(i2_ofr))
|
||||
|
||||
}else{
|
||||
cat("\nAll positions within range"
|
||||
, "\nProceeing with generating requested position MSA seqs...")
|
||||
i2_extract = plot_positions
|
||||
}
|
||||
|
||||
matP2 = wt_interim[i_extract, 1:ncol(wt_interim)]
|
||||
|
||||
dfP2 = data.frame(t(matP2))
|
||||
names(dfP2) = i2_extract
|
||||
cols_to_paste_P2 = names(dfP2)
|
||||
|
||||
dfP2['chosen_seq'] = apply( dfP2[ , cols_to_paste_P2]
|
||||
, 1
|
||||
, paste, sep = ''
|
||||
, collapse = "")
|
||||
|
||||
wt_seq_plot = dfP2$chosen_seq
|
||||
}
|
||||
|
||||
|
||||
######################################
|
||||
# Generating plots for muts and wt
|
||||
#####################################
|
||||
LogoPlotMSAL <- list()
|
||||
|
||||
if (my_logo_col %in% c('clustalx','taylor')) {
|
||||
|
|
@ -78,12 +161,16 @@ LogoPlotMSA <- function(msaSeq_mut
|
|||
#-------------------
|
||||
# Mutant logo plot
|
||||
#-------------------
|
||||
p0 = ggseqlogo(msaSeq_mut
|
||||
#msaSeq_mut$V1
|
||||
p0 = ggseqlogo(msa_seq_plot
|
||||
, facet = "grid"
|
||||
, method = msa_method
|
||||
, col_scheme = my_logo_col
|
||||
, seq_type = 'aa')
|
||||
, seq_type = 'aa') +
|
||||
scale_x_discrete(x_lab
|
||||
, breaks = i_extract
|
||||
, labels = i_extract
|
||||
#, limits = min(i_extract): max(i_extract))
|
||||
, limits = factor(i_extract))
|
||||
|
||||
# further customisation
|
||||
msa_mut_logo_P = p0 + theme(legend.position = leg_pos
|
||||
|
|
@ -111,6 +198,7 @@ LogoPlotMSA <- function(msaSeq_mut
|
|||
|
||||
, plot.background = element_rect(fill = theme_bgc))
|
||||
|
||||
|
||||
cat('\nDone: msa_mut_logo_P')
|
||||
#return(msa_mut_logoP)
|
||||
LogoPlotMSAL[['msa_mut_logoP']] <- msa_mut_logo_P
|
||||
|
|
@ -118,12 +206,16 @@ LogoPlotMSA <- function(msaSeq_mut
|
|||
#---------------------------------
|
||||
# Wild-type MSA: gene_fasta file
|
||||
#---------------------------------
|
||||
p1 = ggseqlogo(msaSeq_wt
|
||||
#msaSeq_wt$V1
|
||||
p1 = ggseqlogo(wt_seq_plot
|
||||
, facet = "grid"
|
||||
, method = msa_method
|
||||
, col_scheme = my_logo_col
|
||||
, seq_type = 'aa')
|
||||
, seq_type = 'aa')+
|
||||
scale_x_discrete(x_lab
|
||||
, breaks = i_extract
|
||||
, labels = i_extract
|
||||
#, limits = min(i_extract): max(i_extract))
|
||||
, limits = factor(i_extract))
|
||||
|
||||
# further customisation
|
||||
msa_wt_logo_P = p1 +
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue