Purely homology-based methods will not work since the statistical likelihood of a random (i.e. false) match to a sequence that short is high. I think Steven Brenner may have a paper on this; from memory 15 amino acids is already too short, and 11 is a hard statistical limit. Something like that.

So your best option is molecular modelling, and the good news is that they like short sequences and actually don't have such a bad success rate (as they do with longer proteins). I think 9 amino acids would be considered a peptide rather than a protein, so you should look for methods for peptides rather than proteins (many of which work by trying to assemble peptide fragments anyway).

Sorry i can't recommend a specific piece of software, but hopefully this should point you in the right direction.

Hi, i suggest u try Gibbs sampler, http://bayesweb.wadsworth.org/gibbs.html , & Emotif http://motif.stanford.edu/emotif-search.html

Calculate your theoretical torsion angles and go from there. http://proteinstructures.com/Structure/Structure/Ramachandran-plot.html

should be able to use a modeller based on these angles

Purely homology-based methods will not work since the statistical likelihood of a random (i.e. false) match to a sequence that short is high. I think Steven Brenner may have a paper on this; from memory 15 amino acids is already too short, and 11 is a hard statistical limit. Something like that.

So your best option is molecular modelling, and the good news is that they like short sequences and actually don't have such a bad success rate (as they do with longer proteins). I think 9 amino acids would be considered a peptide rather than a protein, so you should look for methods for peptides rather than proteins (many of which work by trying to assemble peptide fragments anyway).

Sorry i can't recommend a specific piece of software, but hopefully this should point you in the right direction.

) aminoacids are not sufficient to have a stable fold, so simply your protein cannot have a unique structure

Depending on your sequence you may or may not have a defined structure. Small pepetides such as leucine-enkephalin (Tyr-Gly-Gly-Phe-Leu) does fold into a kind of beta-turn structure, and depending of solution conditions the fraction of the molecules into the folded-conformation may vary

(Behnam BA & Deber CM (1984) Evidence for a folded conformation of methionine- and leucine-enkephalin in a membrane environment. J Biol Chem 259(23):14935-40.

http://www.ncbi.nlm.nih.gov/pubmed/6501322

Aburi M & Smith PE (2002) A conformational analysis of leucine enkephalin as a function of pH. Biopolymers 64(4):177--188.

http://www.ncbi.nlm.nih.gov/pubmed/6501322

).

Try using a disorder prediction site (such as dissemble

http://dis.embl.de/). If you peptide is predicted as dissordered by these servers, then most likely you will not be able to find a defined structure, but a large assemble of possible conformations. If your peptide is predicted as ordered, then try Molecular Dynamics simulations ()MD to search for possible conformations of your peptide. Be sure to run your simulations long-enough, and try harsh techniques, such as simulated-annealing and replica-exchange, to make sure you do a resonable exploration of the conformational space. Please look at this paper, it is very much what you are asking:

http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0027371

You may also try QUARK (http://zhanglab.ccmb.med.umich.edu/QUARK/), this server has shown some success in the prediciton of the folding conformation of small proteins. In this case, do subject the results to MD, to test their conformational stability.

Dear Dushyant,

Alessandro is absolutely correct. The shortest known peptides/proteins with stable folds are around 20 AA long (e.g. apamin defensins etc) but all of them are heavily disulfide connected. This is for a very simple reason that short seuences do not have sufficient interior to have any driving force to stable. MD on short sequences shown that some structures have longer residence times (some hairpins or pseudo helical arrangements) but have stable structure for 9 AA is a contradiction in terms.

Take advice form Rogelio and follow his protocol. The only thing I would add to his advice use Monte Carlo as it may provide better sparse sampling of partition function for the system so you might be able to map relative probabilities of accupancy of different marginally stable conformers. So, the next time when you ask the question do not formulate it in such a radical way: "I have to predict ..." because the task may prove undoable.

Dear Ali,

This thread is answers to Dushyant's question, or comments relating to the answers.

If you want to raise a new question go to your home RG personal page, click on Contributions and on the menu where it reads "Research" you will find

?Questions

Click there and ask your question and be patient.

May I recommend you to narrow down your question as much as possible?

Given the way you wrote it and where you placed it (this thread), your question could be any of these:

- If there is a protein with a sequence of exactly 100 aa?

-How can I find protein sequences with a defined amino acid sequence length?

-Can I model a protein of 100 aa?

or some more...

http://osddlinux.osdd.net/raghava/pepstrmod/