Dear Datamonkey God(s),
When running a dataset through GARD to detect recombination breakpoints, should root sequences be included? For example, I am interested in HIV subtybe B, and use subtype H sequences to root. Will including these more divergent root sequences mask the diversity of the subtype B sequences such that breakpoints are missed?

Thank you!
-Crystal

Hi Crystal,

Rooting is irrelevant for GARD; you can exclude subtype H (I would probably root on D if I were to pick an outgroup). Have you looked at SCUEAL for HIV pol recombination mapping? Or are you looking at different genes?

Sergei

Hi Sergei,

Thank you. I am looking at the env and gag regions. I realize that most people choose subtype D, but most full-genome subtype D sequences appear to have many regions that are B-like when bootscanning methods are employed. Also, loading RDP3 with the same D subtype sequences that are reported in many papers shows that the env regions of D is recombinant with B. I have decided to try using a more distantly related root since B and D are so similar.

-Crystal

Hi Crystal,

Fair enough. Are you looking to identify WHICH sequences are mosaic, or simply find evidence of recombination somewhere?

Sergei

Here is a sample of the similarity between subtype D and B in the env region, in case you are interested. I used REGA, at two mirror sites, and got conflicting results. Also, if you perform multiple runs at the same site, results are often conflicting. Thank again!

---

Hello
I am primarily interested in identifying breakpoints, and then whether a distinct population of sequences included in the dataset tends to include more recombinants. Finally, I am interested in knowing if recombinants tend to cluster together when the parentals are not included.
-Crystal

CrystalH wrote on May 9^th, 2011 at 4:23pm:

Here is a sample of the similarity between subtype D and B in the env region, in case you are interested. I used REGA, at two mirror sites, and got conflicting results. Also, if you perform multiple runs at the same site, results are often conflicting. Thank again!

This is actually very similar to what I have observed with gp160 sequences -- many "pure" references actually have some sort of a recombinant signature.

S.

Hi Crystal,


CrystalH wrote on May 9^th, 2011 at 4:26pm:

Hello I am primarily interested in identifying breakpoints, and then whether a distinct population of sequences included in the dataset tends to include more recombinants. Finally, I am interested in knowing if recombinants tend to cluster together when the parentals are not included. -Crystal

For that you may actually want to use SCUEAL. I can build a reference alignment for gag and env, or you can do it yourself following the instructions at Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login

Sergei

Thank you. I will give it a try myself and if I fail miserably, I will repost. Have a great week!

-Crystal

Hi Sergei,
When I build the reference alignment for my env and gag SCUEAL runs, I was wondering if SCUEAL used complete or partial deletion when identifying breakpoints. I am curious about this, as it has been previously shown that almost all env breakpoints occur in the conserved regions of the env for inter- and intra-subtypes. If SCUEAL uses complete deletion, then I could exclude the V1-V4 variable regions of env. Also, I think this may reduce false signal of breakpoints due to heterotachy. Any thoughts? Thanks!

-Crystal  :-?

Hi Crystal,

Sorry to be dense, but I have no idea what you mean by "partial" and "complete" deletion. I would not place too much weight on any claims about HIV recombination -- it's more about the recombination we can detect.  

Sergei