Repeat copy numbers can be off
Currently, repeat copies are estimated by comparing to average coverage of all contigs.
# copies = floor(repeat contig coverage / avg coverage of large contigs)
However, this can lead to inaccuracies for coverage values close to average coverage. For example:
# copies = floor (99/50) = 1
We should look into a better way of estimating repeat copy numbers.
Also, there could be biases introduced by using FLASH merged reads and unmerged reads in the same assembly.