I agree with Sergii. Another thing to keep in mind about bioinformatic predictions is that they depend on the accuracy of the initial data they are based on. If the initial analyses are wrong, then the prediction will also be wrong.

Another problem to keep in mind is that many proteins in the databases suffer from the "chinese whispers" problem: Their function assignment is based on a prediction which is based on another prediction and so forth. A protein thus removed a few steps from a well characterized one can have quite different capabilities.

One should always keep in mind that bioinformatic software give only a prediction that to any extent corresponds to reality. For example, when you make ab initio prediction of protein spatial structure you can get a probable and thermodinamically good conformation that, nevertheless, has nothing to do with reality. That's why bioinformatic predictions need to be supported by experimental evidence.

One possible solution is to use few bioinformatic programs (servers or standalone soft) based on different algorithms and then compare their predictions. If all programs give similar results then it makes sense to check the protocol of biochemical analysis.

I agree with Sergii. Another thing to keep in mind about bioinformatic predictions is that they depend on the accuracy of the initial data they are based on. If the initial analyses are wrong, then the prediction will also be wrong.

Another problem to keep in mind is that many proteins in the databases suffer from the "chinese whispers" problem: Their function assignment is based on a prediction which is based on another prediction and so forth. A protein thus removed a few steps from a well characterized one can have quite different capabilities.

What type of properties are you looking for? How sure are you that the molecule you look at by bioinformatics is exactly the same as the one you are looking at in the test tube? Posttranslational modifications or non-covalent association (multimerization or association with other proteins, organic cofactors, structurally important bound ions) can significantly affect the behaviour of a protein. Do your simulations or predictions accurately reflect the environment in which the real-life experiments were performed?

Usually the bioinformatics study is based on the PDB structure of a protein reported in RCSB PDB and the coordinates of each atom of amino acids of proteins. Since, during crystallization process many proteins need to be break in parts or cleaved from terminal ends. Such cleaving may lead to slight conformational alterations from its native structure ( in a test tube). Hence, proteins may lead to different characteristic at bioinformatics and biochemical level.

Mr. Kedar Sharma

what kind of biochemical characterization are you interested for analysis...

??

Mr. Tanuj Sharma i m interested in biochemical characterization in terms of substrate specificity.......