Closed lentendu closed 4 years ago
Avoid names and list format for clustering output, there are too large. Also involve modifying the way consensus taxonomy table is build at id step. Start at merge step to store sample count information.
Usage of count_table have been implemented everywhere and thoroughly tested, so fixed
Avoid names and list format for clustering output, there are too large. Also involve modifying the way consensus taxonomy table is build at id step. Start at merge step to store sample count information.