How ariba deal with reads mapped to multiple clusters

[IEkAdN]

Hello,

I would like to use ariba without cd-hit clustering. When two similar reference sequences exist, I guess some sequence reads will be mapped to both of them (in other words, some reads will be mapped to two clusters). In such a case, how does ariba deal with these reads? Does ariba use these reads for local assemblies in both of these two clusters?

[martinghunt]

Yes, reads can be put into more than one cluster. Reads are allocated to every cluster where either read of the pair matches any sequence in that cluster.

[IEkAdN]

Thank you for your quick response. Then I will try ariba without cd-hit clustering.

[mattbawn]

I have also noticed that I get false negatives due to homologs. I have ran with and without '--no_cdhit' but the issue remains. Is there any other recomended manner to overcome this?