msa dashboard

This commit is contained in:
Tanushree Tunstall 2022-09-05 16:05:36 +01:00
parent 88aaf56729
commit 3037d6e3ef
6 changed files with 643 additions and 0 deletions

View file

@ -316,3 +316,344 @@ consurf_colours = c(
, "9" = rgb(0.63,0.16,0.37)
)
if (interactive()){
options(shiny.launch.browser = FALSE) # i am a big girl and can tie my own laces
options(shiny.port = 8000) # don't change the port every time
options(shiny.host = '0.0.0.0') # This means "listen to all addresses on all interfaces"
options(width=120)
options(DT.options = list(scrollX = TRUE))
ui=dashboardPage(skin="purple",
dashboardHeader(title = "Drug/Target Explorer"),
dashboardSidebar(
sidebarMenu( id = "sidebar",
selectInput(
"switch_target",
label="Switch to New Target",
choices = c(
"alr",
"embb",
"gid",
"katg",
"pnca",
"rpob"
),
selected="embb"),
menuItem("LogoP SNP", tabName="LogoP SNP"),
menuItem("Lineage Sample Count", tabName="Lineage Sample Count"),
menuItem("Site SNP count", tabName="Site SNP count"),
menuItem("Stability SNP by site", tabName="Stability SNP by site"),
menuItem("DM OM Plots", tabName="DM OM Plots"),
menuItem("Correlation", tabName="Correlation"),
menuItem("Lineage Distribution", tabName="Lineage Distribution"),
menuItem("Consurf", tabName="Consurf"),
menuItem("LogoP OR", tabName="LogoP OR"),
menuItem("LogoP ED", tabName="LogoP ED"),
menuItem('Stability count', tabName='Stability count'),
# These conditionalPanel()s make extra settings appear in the sidebar when needed
conditionalPanel(
condition="input.sidebar == 'LogoP SNP'",
textInput(
"omit_snp_count",
"Omit SNPs",
value = c(0),
placeholder = "1,3,6"
)
),
# NOTE:
# I *think* we can cheat here slightly and use the min/max from
# merged_df3[['position']] for everything because the various
# dataframes for a given gene/drug combination have the
# same range of positions. May need fixing, especially
# if we get/shrink the imported data files to something
# more reasonable.
conditionalPanel(
condition="
input.sidebar == 'LogoP SNP'||
input.sidebar == 'Stability SNP by site' ||
input.sidebar == 'Consurf' ||
input.sidebar == 'LogoP OR' ||
input.sidebar == 'Site SNP count'",
sliderInput(
"display_position_range"
, "Display Positions"
, min=1, max=150, value=c(1,150) # 150 is just a little less than the smallest pos_count
)
),
conditionalPanel(
condition="input.sidebar == 'LogoP ED'",
sliderInput(
"display_position_full_range"
, "Display Positions"
, min=1, max=150, value=c(1,150)
)
),
conditionalPanel(
condition="
input.sidebar == 'LogoP SNP' ||
input.sidebar == 'LogoP OR' ||
input.sidebar == 'LogoP ED'",
selectInput(
"logoplot_colour_scheme",
label="Logo Plot Colour Scheme",
choices = logoPlotSchemes,
selected="chemistry"
)
),
#conditionalPanel(
# condition="input.sidebar == 'LogoP SNP' || input.sidebar == 'LogoP ED'|| input.sidebar == 'Consurf'",
# numericInput(
# "table_position"
# , "Table Position", value=1
# )
#),
conditionalPanel(
condition="input.sidebar == 'Correlation'",
selectInput(
"corr_method",
label="Correlation Method",
choices = list("spearman",
"pearson",
"kendall"),
selected="spearman"
)
),
conditionalPanel(
condition="input.sidebar == 'Correlation'",
numericInput(
"corr_lig_dist"
, "Ligand Distance Cutoff (Å)", value=1
)
),
conditionalPanel(
condition="input.sidebar == 'Correlation'",
checkboxGroupInput(
"corr_selected",
"Parameters",
choiceNames = c(
"DeepDDG",
"Dynamut2",
"FoldX",
"ConSurf"#,
#"dst_mode"
),
choiceValues = c(
"DeepDDG",
"Dynamut2",
"FoldX",
"ConSurf"#,
#"dst_mode"
),
selected = c(
"DeepDDG",
"Dynamut2",
"FoldX",
"ConSurf"#,
#"dst_mode"
)
)
),
conditionalPanel(
condition="input.sidebar == 'DM OM Plots'",
selectInput(
"dm_om_param",
label="Stability Parameter",
choices = keys(dm_om_map),
selected="SNAP2")
),
# colour_categ
conditionalPanel(
condition="input.sidebar == 'Stability SNP by site'",
selectInput(
"stability_snp_param",
label="Stability Parameter",
choices = stability_boxes_df$stability_type,
selected="Average")
),
conditionalPanel(
condition="input.sidebar == 'Stability SNP by site'",
checkboxInput("reorder_custom_h",
label="Reorder by SNP count",
FALSE)
),
conditionalPanel(
condition="input.sidebar.match(/^Lineage.*/)",
checkboxInput("all_lineages",
label="All Lineages",
FALSE)
),
# an example of how you can match multiple things in frontend JS
conditionalPanel(
condition="input.sidebar == 'LogoP SNP' ||
input.sidebar =='Stability SNP by site' ||
input.sidebar =='Consurf' ||
input.sidebar =='LogoP OR' ||
input.sidebar =='LogoP ED'",
actionButton("clear_ngl",
"Clear Structure")
),
conditionalPanel(
condition="input.sidebar == 'LogoP SNP' ||
input.sidebar =='Stability SNP by site' ||
input.sidebar =='Consurf' ||
input.sidebar =='LogoP OR' ||
input.sidebar =='LogoP ED'",
actionButton("test_ngl",
"Test NGLViewR")
)#,
# downloadButton("save",
# "Download Plot"
# )
# actionButton(
# "reload_target",
# label="Reload Target\nData (slow!)"
# )
)
),
body <- dashboardBody(
tabItems(
tabItem(tabName = "dashboard",
h2("Dashboard tab content")
),
tabItem(tabName = "widgets",
h2("Widgets tab content")
)
),
# creates a 'Conditional Panel' containing a plot object from each of our
# ggplot plot functions (and its associated data frame)
fluidRow(column(width=12,
lapply(plot_functions_df$tab_name,
function(x){
plot_function=plot_functions_df[
plot_functions_df$tab_name==x,
"plot_function"]
plot_df=plot_functions_df[
plot_functions_df$tab_name==x,
"plot_df"]
cat(paste0('\nCreating output: ', x))
generate_conditionalPanel(x, plot_function, plot_df)
}
)
)
),
#### fluidRow()s for "Stability Count" in the sidebar ####
fluidRow(
conditionalPanel(
condition="
input.sidebar == 'LogoP SNP' ||
input.sidebar =='Stability SNP by site' ||
input.sidebar =='Consurf' ||
input.sidebar =='LogoP OR' ||
input.sidebar =='LogoP ED'",
column(NGLVieweROutput("structure"),
width=3
)
),
conditionalPanel(
condition="
input.sidebar == 'LogoP SNP' ||
input.sidebar == 'Stability SNP by site' ||
input.sidebar == 'Site SNP count' ||
input.sidebar == 'Consurf' ||
input.sidebar == 'LogoP OR' ||
input.sidebar == 'LogoP ED'",
column(
DT::dataTableOutput('table'),
width=9
)
)
)
)
)
server = function(input, output) {
observeEvent({
input$combined_model
input$combined_data
input$combined_training_genes
input$score_dropdown
input$resample_dropdown
input$drug_dropdown
input$split_dropdown
},{
combined_model = input$combined_model
selection = input$score_dropdown
resampler = input$resample_dropdown
selected_drug = input$drug_dropdown
selected_split = input$split_dropdown
combined_data = input$combined_data
combined_training_genes = input$combined_training_genes
selected_gene = combo[combo$drug == selected_drug,'gene']
# hide stuff if selected
if(combined_model == "combined") {
#if(combined_model == TRUE) {
hide("split_dropdown")
hide("resample_dropdown")
show("combined_data")
show("combined_training_genes")
filedata = paste0(combined_training_genes,
'genes_logo_skf_BT_',
selected_gene,
'_',
combined_data
)
print(filedata)
print('doing COMBINED plot')
output$plot <- renderPlot(makeplot(loaded_files[[filedata]],
selection,
"none", # always 'none' for combined plot
gene = combo[drug==selected_drug,"gene"],
combined_training_genes = combined_training_genes,
display_combined = TRUE,
)
)
# e.g.
# makeplot(loaded_files$`5genes_logo_skf_BT_pnca_actual`, "MCC", "none" , gene = 'foo', combined_training_genes = '1234', display_combined = TRUE)
} else {
show("split_dropdown")
show("resample_dropdown")
hide("combined_data")
hide("combined_training_genes")
filedata = paste0(combo[drug==selected_drug,"gene"],
'_baselineC_',
selected_split
)
print(filedata)
print("doing GENE plot")
output$plot <- renderPlot(makeplot(loaded_files[[filedata]],
selection,
resampler,
gene = combo[drug==selected_drug,"gene"],
display_combined = FALSE,
)
)
}
# 6genes_logo_skf_BT_gid_complete
# filedata example for combined: 6genes_logo_skf_BT_embb_actual
# 6genes_logo_skf_BT_embb_combined
})
}
app <- shinyApp(ui, server)
runApp(app)
}

View file

@ -98,6 +98,7 @@
<div id="main">
<a href = "drug-target/"><h2>Drug/Gene Target explorer</h2></a>
<a href="ml/"><h2>ML/AI model explorer</h2></a>
<a href="msa/"><h2>Multiple Sequence Alignment explorer</h2></a>
</div>
</div>
</div>

BIN
msa/.global.R.swp Normal file

Binary file not shown.

301
msa/global.R Normal file
View file

@ -0,0 +1,301 @@
library(shinycssloaders)
library(DT)
library(NGLVieweR)
library(hash)
load_target_globals=function(target){
cat(paste0("Reloading Target: ", target))
source(paste0("/srv/shiny-server/git/LSHTM_analysis/config/", target, ".R")) # load per-target config file
invisible(assign(paste0(target, "_merged_df3"), read.csv(paste0("/srv/shiny-server/git/Misc/shiny_dashboard/data/",target,"-merged_df3.csv")), envir = .GlobalEnv))
invisible(assign(paste0(target, "_merged_df2"), read.csv(paste0("/srv/shiny-server/git/Misc/shiny_dashboard/data/",target,"-merged_df2.csv")), envir = .GlobalEnv))
invisible(assign(paste0(target, "_corr_df_m3_f"), read.csv(paste0("/srv/shiny-server/git/Misc/shiny_dashboard/data/",target,"-corr_df_m3_f.csv")), envir = .GlobalEnv))
invisible(assign(paste0(target, "_lin_lf"), read.csv(paste0("/srv/shiny-server/git/Misc/shiny_dashboard/data/",target,"-lin_lf.csv")), envir = .GlobalEnv))
invisible(assign(paste0(target, "_lin_wf"), read.csv(paste0("/srv/shiny-server/git/Misc/shiny_dashboard/data/",target,"-lin_wf.csv")), envir = .GlobalEnv))
lapply(
c(
"duet",
"mcsm_lig",
"foldx",
"deepddg",
"dynamut2",
"consurf",
"snap2",
"provean",
"dist_gen",
"mcsm_ppi2"#,
#"mcsm_na"
), function(x){
wf_filename=paste0("/srv/shiny-server/git/Misc/shiny_dashboard/data/", tolower(gene), "-wf_", x ,".csv")
wf_var=paste0("wf_",x)
if (file.exists(wf_filename)){
invisible(assign(wf_var,read.csv(wf_filename), envir = .GlobalEnv)) # FILTH
}
lf_filename=paste0("/srv/shiny-server/git/Misc/shiny_dashboard/data/", tolower(gene), "-lf_", x ,".csv")
lf_var=paste0(target, "_lf_",x)
if (file.exists(lf_filename)){
invisible(assign(lf_var,read.csv(lf_filename), envir = .GlobalEnv)) # FILTH
}
}
)
}
# populate target-specific *_unified_msa vars
load_msa_global=function(gene){
drug=target_map[[gene]]
in_filename_msa = paste0(tolower(gene), "_msa.csv")
infile_msa = paste0("/srv/shiny-server/git/Data/", drug, "/output/", in_filename_msa)
print(infile_msa)
msa1 = read.csv(infile_msa, header = F)
msa_seq = msa1$V1
infile_fasta = paste0("/srv/shiny-server/git/Data/", drug, "/input/", tolower(gene), "2_f2.fasta")
print(infile_fasta)
msa2 = read.csv(infile_fasta, header = F)
wt_seq = msa2$V1
target_name=paste0(gene, '_unified_msa')
#print(target_name)
invisible(assign(target_name, list(msa_seq = msa_seq, wt_seq = wt_seq), envir = .GlobalEnv))
}
#### Local Functions ####
# Generate a conditionalPanel() for a given graph function and sidebar name combination
generate_conditionalPanel = function(tab_name, plot_function, plot_df){
# e.g.: list("lin_count_bp_diversity", "Lineage diversity count")
cond=paste0("input.sidebar == '", tab_name, "'")
conditionalPanel(condition=cond, box(
title=tab_name
, status = "info"
, width=NULL
, plotOutput(plot_function
, click = "plot_click") %>% withSpinner(color="#0dc5c1")
# , plotOutput(plot_function, click = "plot_click")
)
)
}
# FIXME: passing in the per-plot params is broken
lin_sc=function(x, all_lineages = F, ...){
lf_var = get(paste0(x,"_lin_lf"))
wf_var = get(paste0(x,"_lin_wf"))
cowplot::plot_grid(lin_count_bp_diversity(wf_var, all_lineages, ...), lin_count_bp(lf_var, all_lineages, ...))
}
options(shiny.port = 8000)
options(shiny.host = '0.0.0.0') # This means "listen to all addresses on all interfaces"
options(shiny.launch.browser = FALSE)
options(width=120)
options(DT.options = list(scrollX = TRUE))
################ STATIC GLOBALS ONLY ##############
# never quite sure where "outdir" gets set :-|
# using dataframes instead of lists lets us avoid use of map()
plot_functions_df=data.frame(
tab_name=c(
"LogoP SNP",
"Lineage Sample Count",
"Site SNP count",
"Stability SNP by site",
"DM OM Plots",
"Correlation",
"Lineage Distribution",
"Consurf",
"LogoP OR",
"LogoP ED"
),
plot_function=c(
"LogoPlotSnps",
"lin_sc",
"site_snp_count_bp",
"bp_stability_hmap",
"lf_bp2",
"my_corr_pairs",
"lineage_distP",
"wideP_consurf3",
"LogoPlotCustomH",
"LogoPlotMSA"
),
plot_df=c(
"mutable_df3" ,
"lin_lf",
"mutable_df3",
"merged_df3" ,
"lf_duet" ,
"corr_df_m3_f",
"merged_df2",
"merged_df3",
"merged_df2",
"unified_msa"
)
)
stability_boxes_df=data.frame(
outcome_colname=c("duet_outcome",
"foldx_outcome",
"deepddg_outcome",
"ddg_dynamut2_outcome",
"mcsm_na_outcome",
"mcsm_ppi2_outcome",
"snap2_outcome",
"consurf_outcome",
"avg_stability_outcome"),
stability_type=c(
"DUET",
"FoldX",
"DeepDDG",
"Dynamut2",
"mCSM-NA",
"mCSM-ppi2",
"SNAP2",
"Consurf",
"Average"
),
stability_colname=c(
"duet_scaled",
"foldx_scaled",
"deepddg_scaled",
"ddg_dynamut2_scaled",
"mcsm_na_scaled",
"mcsm_ppi2_scaled",
"snap2_scaled",
"consurf_scaled",
"avg_stability_scaled"
)
)
table_columns = c(
"position",
"mutationinformation",
"sensitivity",
"ligand_distance",
"avg_lig_affinity",
"avg_lig_affinity_outcome",
"avg_stability",
"avg_stability_outcome",
"or_mychisq",
"maf",
"snap2_outcome",
"consurf_outcome",
"provean_outcome",
"rsa",
"kd_values" ,
"rd_values"
)
logoPlotSchemes <- list("chemistry"
, "taylor"
, "hydrophobicity"
, "clustalx")
dm_om_methods = c("DUET ΔΔG"
, "Consurf"
, "Deepddg ΔΔG"
, "Dynamut2 ΔΔG"
, "FoldX ΔΔG"
, "Ligand affinity (log fold change)"
, "mCSM-NA affinity ΔΔG"
, "SNAP2")
dm_om_map = hash(c(
"DUET ΔΔG"
, "Consurf"
, "Deepddg ΔΔG"
, "Dynamut2 ΔΔG"
, "FoldX ΔΔG"
, "Ligand affinity (log fold change)"
, "mCSM-NA affinity ΔΔG"
, "SNAP2"
), c(
"lf_duet"
,"lf_consurf"
,"lf_deepddg"
,"lf_dynamut2"
,"lf_foldx"
,"lf_mcsm_lig"
,"lf_mcsm_na"
,"lf_snap2"
)
)
#### target_map: handy gene/drug mapping hash ####
target_map = hash(
c(
"alr",
"gid",
"embb",
"pnca",
"rpob",
"katg"),
c(
"cycloserine",
"streptomycin",
"ethambutol",
"pyrazinamide",
"rifampicin",
"isoniazid")
)
# load E V E R Y T H I N G
lapply(c(
"alr",
"embb",
"gid",
"katg",
"pnca",
"rpob"
),function(x){
invisible(load_target_globals(x))
invisible(load_msa_global(x)) # turn off to speed up start time at the expense of "LogoP ED"
}
)
consurf_palette1 = c("0" = "yellow2"
, "1" = "cyan1"
, "2" = "steelblue2"
, "3" = "cadetblue2"
, "4" = "paleturquoise2"
, "5" = "thistle3"
, "6" = "thistle2"
, "7" = "plum2"
, "8" = "maroon"
, "9" = "violetred2")
consurf_palette2 = c("0" = "yellow2"
, "1" = "forestgreen"
, "2" = "seagreen3"
, "3" = "palegreen1"
, "4" = "darkseagreen2"
, "5" = "thistle3"
, "6" = "lightpink1"
, "7" = "orchid3"
, "8" = "orchid4"
, "9" = "darkorchid4")
# decreasing levels mess legend
# consurf_colours_LEVEL = c(
# "0" = rgb(1.00,1.00,0.59)
# , "9" = rgb(0.63,0.16,0.37)
# , "8" = rgb(0.94,0.49,0.67)
# , "7" = rgb(0.98,0.78,0.86)
# , "6" = rgb(0.98,0.92,0.96)
# , "5" = rgb(1.00,1.00,1.00)
# , "4" = rgb(0.84,0.94,0.94)
# , "3" = rgb(0.65,0.86,0.90)
# , "2" = rgb(0.29,0.69,0.75)
# , "1" = rgb(0.04,0.49,0.51)
# )
consurf_colours = c(
"0" = rgb(1.00,1.00,0.59)
, "1" = rgb(0.04,0.49,0.51)
, "2" = rgb(0.29,0.69,0.75)
, "3" = rgb(0.65,0.86,0.90)
, "4" = rgb(0.84,0.94,0.94)
, "5" = rgb(1.00,1.00,1.00)
, "6" = rgb(0.98,0.92,0.96)
, "7" = rgb(0.98,0.78,0.86)
, "8" = rgb(0.94,0.49,0.67)
, "9" = rgb(0.63,0.16,0.37)
)

0
msa/server.R Normal file
View file

0
msa/ui.R Normal file
View file