Major update (support for strand-specific sequencing and abundance uncertainty sampling)

[jonassibbesen]

Improvements:

• Added strand-specific sequencing support. See README for more information.
• Added abundance uncertainty estimation using Gibbs sampling. The sampler is initialized on the EM estimate. The number of Gibbs samples can be specified using --num-gibbs-samples (zero is default).
• Added branch and bound like algorithm to infer diplotype probabilities (this is now the default when ploidy is 2)
• Improved output writing to make it more efficient when using multiple threads.
• Updated submodules and added htslib. Please run git submodule update --init --recursive before compiling.

Changes to existing options and outputs:

• Added --output-prefix as a required option. All files written by rpvg will have this prefix (e.g. <output-prefix>.txt).
• Changed probability output option to a flag (--write-probs). The probabilities are now written to <output-prefix>_probs.txt.gz.
• Added compression of probability output and abundance Gibbs samples output (<output-prefix>_probs.txt.gz and <output-prefix>_gibbs.txt.gz).
• Removed --use-exact and added --use-hap-gibbs (combinatorial based inference is now the default).
• Removed the one letter name for many of the less used options.
• Changed Probability header names in haplotyping output to PosterierProbability.

Bug fixes:

• The same seed now gives consistent results.