Align the HIV-1 Gag protein with the Gag protein of Visna virus

Question

Align the HIV-1 Gag protein with the Gag protein of Visna virus

Background Visna virus is a lentivirus causing encephalitis in sheep

The problem is align the follow two protein Gag of viruses HIV-1 (GenBank KU749412.1) and Gag Visna virus (GenBank L06906.1) that are:

"MGARASILRGEKLDTWERIRLRPGGKKHYMLKHLVWASRELDRFALNPGLLETLEGCKQIIKQLQPALQTGTEELRSLFNTVATLYCVHAGIPVKDTKEALDKIEEEQNKCQQKAKQAEAAAAGKISLNYPIVQNLQGQMVHQAISPRTLNAWVKVIEEKAFSPEVIPMFTALSEGATPQDLNTMLNTVGGHQAAMQMLKDTINEEAAEWDRLHPVHAGPVAPGQMREPRGSDIAGTTSTLAEQIAWITGNPPVPVGEIYKRWIILGLNKIVRMYSPXSILDIRQGPKEPFRDYVDRFFKTLRAEQATQDVKNWMTDTLLVQNANPDCKTILRALGPGATLEEMMTACQGVGGPGHKARVLAEAMSQTTNAIMMQRSNFKGPKRTIKCFNCGKEGHLARNCRAPRKKGCWKCGKEGHQMKDCNERQANFLGKIWPSHKGRPGNFLQSRPEPTAPPAESFRFEETTPAPKQEQKDKEPLTSLKSLFGSDPLSQ"   
"MAKQGSKEKKGYPELKEVIKATCKIRVGPGKETLTEGNCLWALKTIDFIFEDLKTEPWTITKMYTVWDRLKGLTPEETSKREFASLQATLACIMCSQMGMKPETVQAAKGIISMKEGLHENKEAKGEKVEQLYPNLEKHREVYPIVNLQAGGRSWKAVESVVFQQLQTVAMQHGLVSEDFERQLAYYATTWTSKDILEVLAMMPGNRAQKELIQGKLNEEAERWVRQNPPGPNVLTVDQIMGVGQTNQQASQANMDQARQICLQWVITALRSVRHMSHRPGNPMLVKQKNTESYEDFIARLLEAIDAEPVTDPIKTYLKVTLSYTNASTDCQKQMDRTLGTRVQQATVEEKMQACRDVGSEGFKMQLLAQALRPQGKAGQKGVNQKCYNCGKPGHLARQCRQGIICHHCGKRGHMQKDCRQKKQQGNNRRGPRVVPSAPPML"

I propose that the better alignment method is the one that maximize the function

f(string1, string2)=(number of letter equal between strign1 and string2)/(common length of strings). 

With that definition I found f(s1,s2)=0.25760 (and 0.2949 between POL protein) for the two strings above.

• What score might others find?
• Is that definition one of biological significative?
• Is there is a better one, what definition for align proteins?
• Are Visna and HIV viruses relatives?

Answer 1

Both are lentiviruses and env is a fast evolving gene. What you are looking at here is the location of indels and this will vary between alignment algorithms.

There will always be an element of error in attempting to align over this genetic distance. It depends what you end goal is, if it is to maximise homology than okay, however optical enhancement of alignments are common if your downstream analysis is phylogenetics. If it is protein structure homology style modelling, where you have the protein structure of HIV env experimentally determined mulitple times you want to use it to help estimate that of the cat lentivirus then optimising a scoring metric is okay.

What you have to understand is there is no 'right' method, in phylogenetics this doesn't matter as much because indels are excluded in the subsequent calculation, so it is less sensative to indel positioning errors and optical enhancement is used. You can also simply exclude regions that appear 'garbled'. For protein homology modeling this is a big problem, because the indels will make drastic changes to the structure prediction. So the only approach here in my opinion is to produce a separate structural prediction for every variation in indel alignment you have obtained. In the latter context getting the alignment right is everything and I suspect thats the output here, because its a classic surface antigen to understand, which will readily crystalise.