BAQLaVA VGB lifestyle prediction

Hello, and thank you for developing and maintaining BAQLaVa.

I am currently working on phage lifestyle classification (temperate, virulent, or unclassified) using BAQLaVa-derived VGBs, and I would be very grateful for your advice on one point.

My main difficulty is that a single VGB can be associated with multiple BAQ reference sequences, and BACPHLIP predictions across those reference sequences are not always consistent. In some cases, different BAQ sequences linked to the same VGB receive different lifestyle predictions.

Because of this, my question is not so much how to interpret an individual BAQ sequence, but rather how to define a representative lifestyle classification at the VGB level when BAQ-level predictions are discordant.

In this situation, could you kindly advise on the most biologically reasonable and defensible way to assign a single lifestyle label to a VGB? For example, would you recommend using a majority-based approach, prioritizing higher-quality reference sequences, or taking a more conservative approach and leaving such VGBs as unclassified?

I would greatly appreciate any guidance you may have. Thank you very much for your time and help.

Seoul national university

min-uk Park

Hi Min-uk,

It is important to remember that the absence of evidence for temperateness/lysogeny is not evidence of lytic-only capacity. This is especially true when considering a group of clustered MAGs, not all of which may be complete genomes.
What we have chosen in our own analysis/assignment of temperateness is to consider a VGB that contains any MAGs/genomes assigned to a status of temperate, as a temperate VGB. Your approach could certainly be enhanced by adding an additional qualifier of completeness if you choose. For example, if using a different rule than our containment-based approach, you probably don’t want an incomplete but virulent-assigned genome to override the status of a complete but temperate-assigned genome within the same VGB.

Hope that guidance helps!
Jordan