In taiawu/dsfworld_package: Visualizes and Analyzes Differential Scanning Fluorimetry Data

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

library(dsfworld)

Minimal version of dye screen processing in an R notebook! For non-dye screen processing, you can try following the worflow in the previous R notebook I sent!

I try to follow tidyverse workflows and conventions as much as possible. The tidyverse is a suite of packages designed to work together to solve common data science problems in coding. Tidyverse documentation: https://www.tidyverse.org/ Tidyverse coding style guide: https://style.tidyverse.org/

I sent you the folder containing the actual current version of the package (which is still in development). I'd recommend just using source() to read in the individual functions, instead of trying to locally install the package. But do what you think is best! FMI on package structure, see the book I followed to build the package https://r-pkgs.org/index.html

Good luck!

I tried to make this notebook easy to re-use. Where I think you'll have to make changes with each screen, i added #USER_UPDATES

Read in libraries

# # these three packaged are used in workflow in this notebook
# library(tidyverse) # actually a suite of packages
# library(fs) # nice filesystem management, used by/with the tidyverse
# library(glue) # nice alternative to paste functions, used to create screen and save names, and used by/with the tidyverse
# 
# # this has all of the actual code for processing dye screens
# library(dsfworld) # you'll likely have to replace this with a series of "source()" calls with the R script names containing the functions. 

Assign some information about your screen, so it's easy to propagate through the following code

# # names / details of screen
# set_protein_name <- "SP115_NHR23LBD" #__USER_UPDATES__
# set_screen_num <- "Exp1236" #__USER_UPDATES__
# set_buffer_used <- "Buffer_P1" #__USER_UPDATES__
# 
# # i like to divide outputs in to different folders. 
# # create those folders here
# # choose the folder names
# plot_folder <- "plots" # save the output plots
# assignments_folder <- "assignments" # save the manual hit assigments df and csv
# int_folder <- "int"  # save the tidied raw data
# # i already have"layout" and "raw" folders in the working directory where this notebook is saved
# 
# # create the directories (or you can do it manually)
# # using the fs library, whih is good and safe. 
# fs::dir_create(plot_folder) 
# fs::dir_create(assignments_folder) 
# fs::dir_create(int_folder)

Assign directories to files that need to be read

# protein_layout_path <- "layout/layout_861_rank_ordered_dye_daughters.xlsx" #__USER_UPDATES__# in this case, the same for protein and buffer
# buffer_layout_path <- "layout/layout_861_rank_ordered_dye_daughters.xlsx" #__USER_UPDATES__# in this case, the same for protein and buffer
# 
# raw_protein_path <- "raw/Exp1152--20210519_SP115_NHR23LBD_dye_screen_861.csv" #__USER_UPDATES__
# raw_buffer_path <- "raw/Exp0724--20191017_OGlcNacTransferase_SP029_buffer_dye_screen_715.csv" #__USER_UPDATES__
# 
# # protein and buffer data don't need the have the same layout. 
# # if the layouts are different, read both in, and pass both to tidied_dye_screen(). See below. 

Step 1: Read in the raw screens and layouts

# ##### unless you want to 
# protein_layout <- # check out documentation--this layout reader is more flexible than the one on the websites
#   read_plate_layout(protein_layout_path)
# 
# buffer_layout <- # check out documentation--this layout reader is more flexible than the one on the websites
#   read_plate_layout(buffer_layout_path)
# 
# raw_protein <- # path to a raw qTower csv
#   read_qtower(raw_protein_path)
# 
# raw_buffer <- # path to a raw qTower csv
#   read_qtower(raw_buffer_path)
# 
# tidied_screen <- # creates / enforces expectations on the data structure of dye screens
#   tidy_dye_screen(.raw_data = raw_protein,
#                   .raw_layout = protein_layout,
#                   .buffer_data = raw_buffer,
#                   .buffer_layout = buffer_layout,
#                   .protein_name = set_protein_name,
#                   .exp_num = set_screen_num,
#                   .buffer_used = set_buffer_used,
#                   
#                   ####  these arguments are passed to label_dye_screen via() ...
#                   #  defaults column names correspond to the echowritr package outputs. 
#                   # see documentation for label_dye_screen() FMI
#                   .dye_col = "compound", #__USER_UPDATES__
#                   .dye_conc_col = "concentration" #__USER_UPDATES__
#                   )

Step 2: save the tidied screen

# write_rds(x = tidied_screen,
#           file = glue::glue("{int_folder}/{set_screen_num}--{glue::glue_collapse(gsub(Sys.Date(), pattern = '-', replacement = ''))}_{set_protein_name}_df_all.rds"))

Make the standard dye screen plot

# full_screen_p <- # takes a minute
#   save_dye_screen_figs(tidied_screen, 
#                        plot_type = "full_screen",
#                        
#                        # passed to saved_stacked_screen() via ...
#                        .save_path = plot_folder)

Call hits This is meant to work with the manual hit selections csv that is created by clicking on the website. In non-interactive versions, for nowm=, just write them as vectors and run this code

If you already have this manual assignments csv saved and read it in via hit_df <- read_csv() You can then bypass the creation and saving to hit_df (next two chunks)

# ####### USER UPDATES THESE ASSIGNMENTS WITH EACH SCREEN
# ## ---- look at the raw data plot to make decisions
# hit_dyes <- # called just a couple, to illustrate example
#   c("A002" ,"A014", "TW408") #__USER_UPDATES__
# 
# sensitive_dyes <- # called just a couple, to illustrate example
#   c( "A003", "L098", "L098") #__USER_UPDATES__

Create the assignment df from the user-defined hits

# hit_df <-
#   # initially, assign every dye "none" 
#   tibble(dye = unique(tidied_screen$dye),
#          assignment = "none") %>%
#   
#   mutate(assignment = if_else(dye %in% hit_dyes, # amend if "hit"
#                               true = "hit",
#                               false = assignment),
#          assignment = if_else(dye %in% sensitive_dyes, # amend if "sensitive
#                               true = "sensitive",
#                               false = assignment))
# 
# hit_df # see requirements for hit_df in the documentation ofr save_dye_screen_figs()
# 
# assignment_df_name <- glue::glue("{set_screen_num}--{glue::glue_collapse(gsub(Sys.Date(), pattern = '-', replacement = ''))}_{set_protein_name}_manual_assignments")
# 
# # save as an rds
# write_rds(x = hit_df, 
#           file = glue::glue("{assignments_folder}/{assignment_df_name}.rds"))
# 
# # save as a csv
# write_csv(x = hit_df, 
#           file = glue::glue("{assignments_folder}/{assignment_df_name}.csv"))

Save the by-hit dye figures

# ##--- plots the same data as the original plot, but broken into sub-plots by assignment
# full_stacked_screen_p <- 
#    # notice this is the same function used to save the original screen
#   save_dye_screen_figs(tidied_screen, 
#                        plot_type = "divided", # this argument determines which plot is output
#                        hits = hit_df, # only "full_screen" does not require this argument 
#                        # passed to saved_stacked_screen() via ...
#                        .save_path = "plots")
# # sometimes prints this warning, which is nbd. Something about the sub-plot alignment probably need a minor fix but the output plot is fine.
# # `Graphs cannot be vertically aligned unless the axis parameter is set. Placing graphs unaligned.`
# 
# ##--- plots the hits and sensitives, this time, faceting by channel.
# full_stacked_screen_p <- 
#    # notice this is the same function used to save the original screen
#   save_dye_screen_figs(tidied_screen, 
#                        plot_type = "hits_by_channel", # this argument determines which plot is output
#                        hits = hit_df, # only "full_screen" does not require this argument 
#                        # passed to saved_stacked_screen() via ...
#                        .save_path = "plots",
#                        
#                        # increasing `.default_title_height` over it's default (0.5) adds some height to the figure if it's too short, and clipping axes. 
#                        # works like this for all save_dye_screen_figs plots.
#                        .default_title_height = 2 
#                        )

taiawu/dsfworld_package documentation built on June 18, 2024, 5:39 a.m.