Skip to content

Latest commit

 

History

History
58 lines (56 loc) · 3.02 KB

TODO.org

File metadata and controls

58 lines (56 loc) · 3.02 KB
Raw

Code clean-up [7/7]

  • [X] break up flowHist into smaller reusable chunks, to facilitate stepping through the init process, and redoing individual steps
  • [X] use consistent names for flowHist objects: pick one of fh, self etc and replace all the others.
  • [X] use consistent syntax - pass and return complete flowHist objects, rather than individual slots
  • [X] incorporate flowHist slots into flowFrame objects?
  • [X] use proper named lists for the components
  • [X] consolidate model component code Note: adding a new model component requires:
    • adding a the component function in models.R (i.e., brB), including a “col” attribute for plotting
    • add an if statement to add the component in flowHist() in flowHist.R
    • adding an if statement to add initial values in flowInit() in models.R
    • add appropriate colors to coltab in plot.flowHist() in flowHist.R

    Component colors should be are now stored as attributes of the components themselves. Can initial value functions also be added as attributes? Yes, and also functions to check if a component should be added to a FlowHist object. All self contained now!

  • [X] update vignettes and examples to use flowPloidyData package to access sample files
  • [ ] The gate plot is updating to earlier when clicking ‘set gate’. It needs to wait until after the gate is set. As is, it ends up out of sync with the applied gate. Need to study the way reactive functions are processed and arrange for the correct order.

Features [8/9]

  • [X] Need to provide an interface for visualizing an existing gate in the browseFlowHist UI.
  • [X] Gating
  • [ ] save and restore parameter estimates from nls
  • [X] S-phase After considering broadened trapezoids and broadened rectangles, and looking at the resulting parameter values and the influence of the different parameters, a simplified “broadened rectangle” seems appropriate:
    • the absolute value for the height of the rectangle is very small relative to other model components - often < 10
    • varying the sd makes a trivial change to the RCS: i.e., varying sd from 1 to 100 moved RCS from 1.271 to 1.306.

    Consequently, I’m using a fixed sd of 1 for the broadened rectangle, and avoiding the additional complexity/parameters that would be involved with the broadened trapezoid with varying sd and starting and ending heights.

  • [X] linearity Linearity parameter is now added and functioning properly in the linearity branch. Not sure how to deal with this just yet. The sample data ranges from 1.92 – 2.07. How much is acceptable? Should it be used to ‘correct’ the mean ratios?
  • [X] initial peak detection with smoothing using kza::kz (or FFT?) Different smoothing algorithms just shift the trade-offs between accuracy and noise in the data. Tried kza:kz, kernalapply, rollapply.
  • [X] export results as a table for saving to disk
  • [X] batch processing
  • [X] switch to nlsLM for regressions