search

kemball / karkadann

The Unicorn Progect 2
0 stars 0 forks source link

cluster deduplication #3

Closed kemball closed 8 years ago

kemball commented 8 years ago

Cluster deduplication is explicitly wrong.

I found a siderophore cluster with 26 genes and only 14 distinct, which implies all kinds of stuff about cluster combining. Mostly that it doesn't work, which is not good.

kemball commented 8 years ago

Turns out start and end values were being stored as strings, not integers. This enforced a lexicographic ordering(alphabetical) not numerical. This made all the inter-cluster logic insane. It's fixed now.