search

datarail / gr_metrics

15 stars 12 forks source link

Added code for GR statistics for GR_s and GR_d #38

Closed yunguan-wang closed 5 years ago

smkartik commented 5 years ago
  1. Include conditions under which dead cell count estimate is corrected. Refer to lines 152 to 213 in https://github.com/datarail/DrugResponse/blob/1d0fd1898e797b74f611eb61dbd6926ac715bb83/MATLAB/import_columbus/Process_CellCountData2.m
  2. Leave out converting time units from hours to days as it pertains to a use case we are not considering in this context.
yunguan-wang commented 5 years ago
  1. Include conditions under which dead cell count estimate is corrected. Refer to lines 152 to 213 in https://github.com/datarail/DrugResponse/blob/1d0fd1898e797b74f611eb61dbd6926ac715bb83/MATLAB/import_columbus/Process_CellCountData2.m
  2. Leave out converting time units from hours to days as it pertains to a use case we are not considering in this context.

Fixed above issues. I left the time conversion part intact but modified its default behavior to outputting GR_d (hour). Please see now commit.

smkartik commented 5 years ago
  1. Include conditions under which dead cell count estimate is corrected. Refer to lines 152 to 213 in https://github.com/datarail/DrugResponse/blob/1d0fd1898e797b74f611eb61dbd6926ac715bb83/MATLAB/import_columbus/Process_CellCountData2.m
  2. Leave out converting time units from hours to days as it pertains to a use case we are not considering in this context.

Fixed above issues. I left the time conversion part intact but modified its default behavior to outputting GR_d (hour). Please see now commit.

  1. The condition described in lines 199-213 are still missing from the implementation.
  2. The purpose of the code is to compute gr_static and toxic. Leave out the time conversion as it is irrelevant to that main purpose. We should keep code as simple as required. I don't see how anyone would want to provide a table with time in hours and then pass an argument t_units=day to this function in order to convert it to days. If in the future we have need of the conversion, there will a better place to implement it.
yunguan-wang commented 5 years ago
  1. Include conditions under which dead cell count estimate is corrected. Refer to lines 152 to 213 in https://github.com/datarail/DrugResponse/blob/1d0fd1898e797b74f611eb61dbd6926ac715bb83/MATLAB/import_columbus/Process_CellCountData2.m
  2. Leave out converting time units from hours to days as it pertains to a use case we are not considering in this context.

Fixed above issues. I left the time conversion part intact but modified its default behavior to outputting GR_d (hour). Please see now commit.

  1. The condition described in lines 199-213 are still missing from the implementation.
  2. The purpose of the code is to compute gr_static and toxic. Leave out the time conversion as it is irrelevant to that main purpose. We should keep code as simple as required. I don't see how anyone would want to provide a table with time in hours and then pass an argument t_units=day to this function in order to convert it to days. If in the future we have need of the conversion, there will a better place to implement it.

For line 199~213, I will contact Marc for clarification for the statement below and implement the changes. Dratio_gr = max(t_c.Deadcount(Eqidx) - t_c.Day0DeadCnt(Eqidx),1)* (1./b + diag(( repmat((-1)*(a-b),1,35).^repmat(1:35,length(a),1) ) * (repmat(b,1,35).*repmat(1:35,length(b),1))')) As for now, it does not affect the output and will only raise an issue when the Ctrl cell count equals the T0 cell count exactly. For the time conversion, I can remove it from the code once everything else is working but not before since it is essential to benchmark my code against the original output file from Marc's Matlab code.

jmuhlich commented 5 years ago

Superseded by #40.