Skip to content

notes_plate-reader.md

Latest commit

 

History

History
63 lines (49 loc) · 5 KB

notes_plate-reader.md

File metadata and controls

63 lines (49 loc) · 5 KB

Notes of Plate reader analysis. tags : #notes

General

Features

  • Need to include the units of the fluorescence data (a.u or MEFL) in the csv file that is saved
  • MEFL automation : get old calibrants from file in processed/..csv?. Stand-in values since we're using fixed gain of 90 for most readings now a days ; ideally add fresh calibrants if data is important..
    • Change switch to not check for native calibrants / add another switch for getting calibrants: in 2-read_multiple_Grids..R : MEFL_normalization <<- ..
    • (future implement) : Look for Gain in the first col of fl$sheet and read the 6th col of this .. need for all fluor; it's laborious
  • (later) Might be a good idea to include the row and column indices in the processed.data for cross-checking/data provenance?
  • (ignore, can't understand) Extrapolation of regex in metadata : mixing some regex with some letters separated by spaces doesn't work, since the spaces are just skipped :( :( Not enough context to recall what this was about)

Dose_reponse plots

Work on this is going on in branch : incorporate_dose_response -- need to merge basic plots with this before adding any work : 18-April-24

  • Build safety into the hill fit using purrr::safely() as in qPCR RAM stability
  • show hill coeficients on plot in plot_dose_response() in 4-plotting_fns.R : .data$fit[[1]]$coefficients gives a named vector!
  • Do a try..catch workflow to choose between hill fits and just connecting the mean points with a line? note from commit: fb96971
  • Use insurance for hill fitting failure. Can use functions purrr::safely() or possibly(). Issues explained below -
    • guide for error catching helpers from purrr : aosmith blog
    • possibly() and safely() work with functions as input, but the hill fitting function : drc::drm() value is a drc object and not functions
    • Contrast with stats::lm : lm returns an object of class : "lm"; which is a list containing stuff.. - probably recognized by the above functions
  • How to retain the name of the inducer and units into the plots? (ex: AHL_nM)

Generalization of markdown plots

Need to incorporate the inducer column in the basic_plots pipeline

  • Make an if loop for when Inducer %in% colnames(..) make a dose response plot borrowed from dose_response_. dose_response_plots.Rmd
  • Run the regular basic_plots script (in a separate child .Rmd file as the else clause) : Will do once the above is settled.
    • Sould these both be called from the .R script to avoid a dummy parent .Rmd file? The data output from this file useful generally ; so keep it?

Analysis

  • Implement a neat shortcut to get metadata that are same for the whole row/column with an 'extrapolation' function
  • Implement a OD subtraction of baseline PBS/LB. Check if this keeps the linearity.. Actually need to analyze the serial dilution for OD linearity first.. Noticed that the OD of culture and 5x concentrated PBS slurry is comparable OD, this is concerning :: S037 data
  • Incorporate the MEFL normalizations if found and update legend accordingly

Short steps

  • Check how the data is read in current script - Is Inducer read in properly?
  • port code from other branch : dose_response_.

Bugs

  • Fix the geom_text layer in the interactive plot (RFP_mean - remove label/shorten, Round decimals? / Samples = 0.9 -- what is this?). This works fine in regular plot, only poor in ggplotly. Example : ![[Pasted image 20230318160713.png]]
  • Extrapolation of regex in metadata : There is a silent bug that is missing a few links from sample name to inducer and links the wrong.

17/1/22

Script: dose_response_plots.Rmd

  • hill_fit.SS() : trying out 2 self starter functions : convergence failed..(with GFP data of SS.. hmm, GFP has no trend here, look at mCherry)
  • Need some way to figure out which data needs to be fit and which ignored due to convergence issues.. example: negative controls, PBS etc.

Continuous_growth_curve

features

  • Add MEFL feature to kinetic read input/processing

  • Add growth rate for OD only and GFP/OD data -- Shyam is interested

  • Generalize the plotting from only GFP to other fluorescences : Like 106 and more in run_kinetic_reads.R, with the logic of 67 and 71 in static_plate_reader..Rmd in branch:incorporate_dose_response

  • Merge incorporate dose response into main, looks like too many things are happening in this branch so next time split into different branches for each small task

Progress

  • Fixed bug for plot_kinetic_raw and ribbon.summary functions Time (hr) changed to Time(h)
  • Updated readme to have an image. Jiwoo is testing the scripts 😊
  • Pushed changes to basic_plots branch