In article <20aa3c9a.0402262039.27a0d39c at posting.google.com>, bigc wrote:
> I'm a newbee to bioinfomatics. It seems that via the secondary
> structure we can roughly predict the protein's 3D structure. Then is
> there free software (better open source) can display a file like this:
> 70 80 90 100 110 120
> ....:....x....:....x....:....x....:....x....:....x....:....x
> VTVQHKKLRTPLNYILLNLAVADLFMVLGGFTSTLYTSLHGYFVFGPTGCNLEGFFATLG
> 1236136_1236137
> EEECCCCCCCCCCCCHHHHHHHCCCCCCCCCCCCEECEEEEECCCCCCCCCCCCCCCCCC BPS
> EEEHHHHHHCCCCEEEHHHHHHHHEHEECCECCCEECCECCEEEECCCCCCCCCHHHCHC D_R
> EEEECCCCCCCHHHHHHHHHHHHHEEEECCCCCEEEECEEEEEEECCCCCCCCEEECCCC DSC
> HHHHHCCCCCCHHHHHHHHHHHHHHHHCCCCCCEEEECCCEEEEECCCCCCCCCEEEECC GGR
> HHHHHHHCCCHHHHHHHHHHHHHHHEEEECCCEEEEEEEEEEEEECCCCCCCCEEEECHC GOR
> ECCCCCCCCCCCCCCCHHHHHHHHEEEECCCCCEEEECCCCEEEECCCCCCCCCCCCCCC H_K
> EEEEECCCCCCCEEEEEEEEECEEEEEEEEEEEEEEEEEEEEEEEECCCCCCCEEECCCC K_S
> EEECCCCCCCCCCCCHHHHHHHHHEEEECCCCCEEEECEEEEEEECCCCCCCCCEECCCC JOI
>> In 3D?
It's not clear what bigc is asking for here.
Does he want to make a 3D prediction from a number of inconsistent
secondary structure predictions? If so, good luck! No one has yet
come up with a method for doing that. I don't recognize most of the
secondary structure predictors he is using, but those I do recognize
are a couple of software generations back and not the best predictors
any more (if they ever were).
Does he want to take a known 3D structure and color it with secondary
structure predictions? That's easy---I wrote a short perl script that
takes my predictions and uses them produce rasmol commands for
selecting and coloring residues. Since most of the script is parsing
the specific format used by the secondary structure predictor, there
is not much point to sharing the script---you'd need a new one for
each different format anyway.
Does he want to use the secondary structure in some way to guide fold
recognition? Lots of different approaches have been used for that,
and it does seem to be fairly successful. Look at any of the
fold-recognition methods that did will in CASP4 or CASP5---they almost
all used some form of secondary-structure prediction to influence the
fold recognition. It's generally better to work with a probability
vector over the secondary structure alphabet though, as the confidence
of the prediction varies enormously from position to position.
--
Kevin Karplus karplus at soe.ucsc.eduhttp://www.soe.ucsc.edu/~karplus
life member (LAB, Adventure Cycling, American Youth Hostels)
Effective Cycling Instructor #218-ck (lapsed)
Professor of Biomolecular Engineering, University of California, Santa Cruz
Undergraduate and Graduate Director, Bioinformatics
Affiliations for identification only.
