Confidence thresholds

This release sees some helpful additions:

• 95% confidence intervals for each of the variants and the cMAF (combined minor allele frequency), as well as the confidence threshold for each amplicon. The confidence threshold is the median value for the high end of the 95% confidence interval of the variants of an amplicon, which is basically the background level. If the low end of the 95% confidence interval for a variant is above this confidence threshold, we can be fairly certain that the variant is real. There are probably many real variants below that threshold as well, but this is a conservative filter to get a high-confident set of variants.
• Added feature to count up the bases before and after merging reads with PrimerID, so we can compute the overall background noise reduction
• Now combine_linkage_values.pl actually does merge files! The merge_replicates sub was just a shell before. It computes a combined p-value using Fishers method.
• Updated compare_variant_frequency script to work with the new output format for convert_reads script. Added additional columns to show whether the variant in a comparison pass the confidence threshold for the amplicon, and the log2 ratio of change in cMAF between the two samples being compared. I changed the filter for this script from just using adjusted p-value < 0.05 by adding filter for absolute log2Ratio_cMAF > 1, and requiring that at least one of the variants in the comparison is above the confidence threshold.

And a few bug fixes:

• Modified calculate_linkage script to compute the p-value outside of the Parallel loop. Much faster and no more Statistics::R errors for jobs with lots (e.g., 500+) comparisons.
• Modified merge_tally script to not print extremely low coverage variants (usually beyond the reference).