TODO

Author: Susana Posada Cespedes
-> User option to indicate if module table contains module names and variant names
-> Add R and functions subfolders to the search path (when modseq_main.R is run), or add functions to an R package
-> Use fastqsampler
-> fastq file after paired-end, read mapping and filtering
-> Simulated data: use read simulators, such as GemSim 
-> Consider using beta-binomial test of strand bias to account for systematic sequencing errors
-> Test run (function to subsample reads available, but not incorporated in an automatized way)
-> Add test cases (ideally testing all possible options) and provide way to run it from the command line

Run2_bwa
-> Remove hyphen separating pairs of unmerged reads (Cases in which pandaseq failed). If the hyphen is not removed, the sam file is not output properly (CIGAR and read sequence are of different lenth - truncated file)

Run3
-> Test ngshmmalign: in order to map reads to most likely module combination use hamming distance (note that the position with respect to reference is defined by the msa)

Run3_gls
-> Verbose level fo run3
-> Map.mode using pairwise alignment
-> Mismatch allowance per module (at the moment perfect match or max. allowance per search round)
-> Mclapply -- Problem when running in euler 

Run3_bwa
-> Split read mapping: module table w.o empty modules and then run bwa on unaligned reads including module combinations with empty modules 

Run4_bwa
-> Exclude duplicates in the distribution (flags 1024, 1040)?
-> Plot of the progression of the alignment filtering
-> Plot of the distribution of start position and end position (combined with automatic option to choose the threshold)

Run5_bwa
-> For reads that are realigned, the MD is not provided (it is used for the pileup model) 
-> Rcpp errror:
qError in dyn.load("/tmp/Rtmp5Qowjy/sourcecpp_5ff840502215/sourceCpp_1.so") :
  unable to load shared object '/tmp/Rtmp5Qowjy/sourcecpp_5ff840502215/sourceCpp_1.so':
  /tmp/Rtmp5Qowjy/sourcecpp_5ff840502215/sourceCpp_1.so: failed to map segment from shared object: Operation not permitted
Calls: source ... sourceCpp -> source -> withVisible -> eval -> eval -> dyn.load

-> Susana: Docker (option for running modseq in a platform compatible way)
RStudio server: To share files and folders between the docker image and host OS one can use the -v option. This acts much like running RStudio in the working directory.
https://github.com/rocker-org/rocker/wiki/Using-the-RStudio-image

-> Steven: Shiny server (option for the visualization)
Asynchronous exchange for long-running calculations in R
https://hub.docker.com/r/rocker/shiny/

-> Define the decision making process (plots and statistics)