The improvement effect of extend

[zhangwenda0518]

Hello, thank you for developing such a comprehensive virus analysis pipeline.

I am very interested in the extend process you created and would like to ask further about its performance. I applied the Leiden clustering method to my dataset, yet I found that using extend in some cases actually shortened the cluster references. Could you please advise on how to improve the performance of extend?

Thank you. This is a thorough and practical tool.

[UriNeri]

Hi, do you mind sharing an example (inputs, command used, and outputs)?

Note that the extend command is still experimental, for now it was only tested on some private data so I can't offer a good performance comparison for it. What I would recommend is trying it on small or selected sequences and at least inspecting the results manually.
The extend command internally tries to avoid duplication and self loops. Because the command largely examines the terminal edges for potential alignments, if the original sequence had repeats, I think it might be possible that the result would actually be a trimmed version.
Alternatively, I can't remember if I made the representative selection to be longest cluster member (explicitly) for all clustering selection, so that may cause weird and more inconsistent behavior between the different clustering options.
Edit - I closed the issue by mistake, reopening