Hi,

I have some full-length HIV-1 A1/D recombinant genomes that I would like to map using SCUEAL.  However, I'm not sure what is the best method for creating a reference alignment in the program.  The SCUEAL readme file notes that the 'build reference alignment' method might not be the best method to use if gappy genes such as env are involved, due to the repeated alingments that occur when building the reference alignment.  Is it better for me to create my own reference alignment and then upload it using the 'make reference alignment' script?

Thanks in advance.

Geraldine.

Hi Geraldine,

For full length alignments it is definitely NOT a good idea to use BuildReferenceAlignment.bf, but rather MakeReferenceAlignment.bf
The two pitfalls with the latter approach are worth noting:

• Ensure that the hand-made alignment is in-frame; SCUEAL works best when query sequences can be codon-aligned to the reference
  
• Run the reference alignment through GARD to ensure that none of the subtype references are between- or within- subtype recombinants themselves as this would bias subtyping.
  

Sergei

Brilliant.   Thanks so much!

A further question: when I create full-length alignments incolving several subtypes, there are often regions (e.g. the V3 loop) that are essentially unalignable.  I usually strip these regions from the final alignment so I don't create errors.  Is this also a good approach when making a reference alignment for use in SCUEAL?  Will the missing region cause problems with the algorithm?

Thanks again!

Geraldine

Dear Geraldine,

Stripping poorly aligned regions should work; in my experience, SCUEAL works best on single genes -- the evolutionary process along the entire viral genome is quite complex and is not well modeled by a single tree/model combination. It would probably be best to make an ENV alignment only (sans V3, as you say), and see how SCUEAL performs on it first.

Sergei


Geraldine wrote on Aug 23^rd, 2010 at 4:42am:

A further question: when I create full-length alignments incolving several subtypes, there are often regions (e.g. the V3 loop) that are essentially unalignable.  I usually strip these regions from the final alignment so I don't create errors.  Is this also a good approach when making a reference alignment for use in SCUEAL?  Will the missing region cause problems with the algorithm? Thanks again! Geraldine