Detailed steps for analysis:

###### preparation ###### 

filelist_bdt is a good summary of all files

1. ./convert_bdt.pl
   Calculate the new BDT scores
   Remove events used in the training ...
   output is stored in ./checkout_rootfiles_correct_bdt

not needed if using most recent WCP to produce the samples

##### CV files ##### 

1. ./convert_cv.pl
   filelist_cv as control
   For CV rootfiles, elimate duplicated events
   Also remove failed run/subrun, if the failed percentage is larger than 20% (default)
   output is stored  ./processed_checkout_rootfiles
   branches status are set to ZERO + required variables 



2. For data, need to run the pot tool to get POTs
./summarize_pot.pl 
<-- input database files in pot_counting [from Haiwang]
--> output pot_bnb.txt and pot_extbnbn.txt to be consumed by app pot_counting 
./bin/pot_countng_mc #mc_file
./bin/pot_counting #bnb_file                   --> get the POT for BNB ...
./bin/pot_counting #bnb_file  #extbnb_file -m2 --> get the POT for EXTBNB ...
change ext_POT in ./configuration/cv_input.txt

3 ./convert_histo.pl
  For CV rootfiles, store the histograms in ./hist_rootfiles ...
  cv_input.txt: files, and external POT, etc. (reading breaks if "-1" [end line] shows up)
    ATTENTION:  BNB data filetype has to be "5"
  file_ch.txt: breakdown cuts, other cuts, etc for each file, not file each channel
  cov_input.txt: channel, x-axis variable, range, etc
  convert_checkout_hist needs to activiate branch in various trees before cuts.h or anything else can use it !!!! 
  be careful LowE sample which only has <400 MeV events and the POT counting is only for lowEnu part. Need to double "add_cut" (+LowEnu) to account for this LowE sample.
  currently (as of Nov 6) dirt sample CV use old GENIE tune results because it has a large stat.

4. ./bin/merge_hist -r0 -l1  (form CV comparison with standard error propogation)
   cv file is stored in merge.root ...
   TLee need LEEx1 -l1 as input or modify TLee source code to enable LEE portion

4#. plot_hist based off merge_hist + truth breakdown + total uncertainty + elegant plotting
need to run step 3 with breakdown file_ch.txt
Do not do this for mc_stat
Be sure the channal name in cov_input.txt has a clean _ext_, _dirt_, _LEE_ sections. The other breakdown keywords correspond to add_cut names

5. ./run_mc_stat.pl (for MC statistics)
    output are stored inside mc_stat/
for Bayesian procedure, better use fewer subchannel (no breakdown) otherwise may fail to do convolution.


###### detector systematic uncertainties ######

No need to run conver_cv.
soft link from processed_checkout_rootfiles to original files
6. ./merge_det.pl
    Merge CV+DetVar into one for each of the 10 sources
    output are stored inside ./hist_rootfiles/DetVar/
    add new variables
  add new variables in merge_xxx.cxx and master_cov_matrix.cxx and mcm_1.h

7. ./run_det_sys.pl
    Running the detector systematics with bootstrapping
    output are stored inside ./hist_rootfiles/DetVar/
One could disable some channels (low statistic) in det_cov_matrix.cxx
10 min

##### Xs and Flux uncertainties ######

8 ./merge_weight.pl
  Merge the CV + weight files into one for each of 17 types of weights 
  output are stored inside ./processed_checkout_rootfiles/*/
  20 CPU*hour?
  add new variables in merge_xxx.cxx and master_cov_matrix.cxx and mcm_1.h

9. ./run_xf_sys.pl
  Running the xs and flux systematics with reweighting
  output are stored inside ./hist_rootfiles/XsFlux/
One could disable some channels (low statistic) in xf_cov_matrix.cxx



#### plotting code ###

./bin/merge_hist -r0  (rerun ...)

./bin/read_TLee_v20

./bin/plot_hist -r0 -l0 -c1 -e3 -sfile_collapsed_covariance_matrix.root