What Is The Concept Of Secondary Structure Prediction By Using Ncbi Blast, Hhpred And Jpred?

Question

[Pont]

Please kindly let me know, the concepts of secondary structure prediction by using NCBI BLAST, HHpred and Jpred. Are they any links. My teacher gave me one sequence and by using these program, I need to search the domain. When will start and when will end. Firstly I used NCBI, secondly I used HHpred and then I use Jpred.I found PDZ doamin and checked the secondary structure at HHpred and Jpred , I got the same results. So I think this is the my target domain. I answer to my teacher. But my teacher asked me again, why do u think this is your PDZ domain? What is the evidence? How about the other you not think domain? etc., so many how??? and why??? . Actually I am not too familier with this , so I can not answer. But I have to gave answer to my teacher on tomorrow. So please share me your knowledge, suggestionas and ideas.

Thank you Pont

Answer 1

[Chris Evelo]

Yes sure. Please start by reading the help pages at:

• NCBI Blast
• HHpred
• Jpred

After that you might have more specific questions.

[Pont]

Thank you for your kind suggestions.I will try to read.

Answer 2

[Lars Juhl Jensen]

Adding to Chris' good suggestion, I suggest the following additional reading for extra credit:

• http://www.ncbi.nlm.nih.gov/pubmed/18463136
• http://www.ncbi.nlm.nih.gov/pubmed/19626712

I also might add that NCBI BLAST did not do secondary structure prediction last time I checked.

[Lars Juhl Jensen]

Yes, I saw that one. Still waiting for them to add RTFM ;-)

[Pont]

Thank you, I will try.

[Chris Evelo]

Did you know the Oxford English Dictionary added LOL to its pages recently? http://www.wired.com/underwire/2011/03/omg-oxford-english-dictionary/ (sorry couldn't resist)/

Answer 3

[Jan Kosinski]

Come on lets help, Pont needs to give the answer tomorrow ;-)

HHPred should be sufficient to answer your question, no need to run BLAST or Jpred (but don't tell this to your teacher if he said you do need that programs ;-)

1. Run HHPred with your sequence. As a database (Search options -> Select HMM databases) select a database named "cdd..."
2. You will get a list of alignments to known domains with scores.
3. "The evidence" are high probability ("Prob" column) and low E-value. If you get Prob > 80-90, and E-value lower than 10E-03 you are almost sure that your domain is there. If it is school exercise, I guess you do get such scores for at least any domain.

Remember that you can have multiple domains in your protein. And they may overlap. If they all get high scores - all may be valid.

And read section 2.7 "How can I verify if a database match is homologous?" from ftp://ftp.tuebingen.mpg.de/pub/protevo/HHsearch/HHsearch1.5.01/HHsearch-guide.pdf

After that, read all the above papers ;-)

[Pont]

Thank you for your help.