OK

Please let me get any suggestion or clue to search ....

It all depends on what type of protein you have. Make sure you subtract the ellipticity of the sample protein from the buffer. For example, I did a run of BSA and got the secondary structure prediction as 68.44% (alpha-helix) and 8.56% (beta strand) which is kind of incorrect as it should have reported 100% for BSA (BSA has no beta strands).

The output from K2D3 will suggest how much error does the data have and you have to rely on that and make some decisions. If it is an unknown protein, make sure to perform other analytical tools to help contribute to the identity.

thank you for your reply. please let me know the other analytical tools. can you please suggest me.

Other analytical techniques such as FTIR and analyzing the Amide-I band or Dynamic Light Scattering in addition to DSC. I have not performed these but you can try.

I am sorry I wasn't clear on the unknown protein part. By other analytical tools I meant combining the data obtained from say LC-MS and digesting it with proteolytic enzyme such as Trypsin and mapping the peptides and hope that you can compare it to a database if one exists for that.

This is an old question, but I guess that the matter is still relevant for those who start to work with a virtually unknown protein (but with known sequence).

Obtain the so-called FASTA form and start with prediction methods:

-First secondary prediction methods like PsiPred. It is common practise to use at least two different methods. See for a little comparison between a number of the most frequently used methods:

Article The role and significance of potential lipid-binding regions...

-A 3D prediction might be considered, I-TASSER is one of the best methods for this purpose (and result in a pdb file which can be used for further analysis). See for an example:

Article Identification of potential lipid binding regions in cereal ...

-Possible alpha-helix regions can be further analysed by Heliquest and renders helical wheel plots and all sorts of physicochemical data:

Article HELIQUEST: A web server to screen sequences with specific α-...

One need to keep in mind that predictions have a certain degree of uncertainty (or put in more positive words an accuracy of round 90%), so intrinsically it is possible that your protein is an exception and the predictions are not entirely accurate. Therefor experimental confirmation is necessary for example by circular dichroism. However, ‘simply’ put protein in buffer and measure might not always render relevant data. Possible influence of difference in variables like pH, ionic strength, temperature etc. need to be checked. Also, for example the absence or presence of anionic phospholipids can dramatically affect the secondary structure, see for example:

Article Anionic phospholipids are essential for α-helix formation of...

Chapter Characterization of helical structures in gelatin networks a...

In the ideal world the predictions and experimental data are in line (at least in order of magnitude). If not either the CD analysis (all sorts of methods are available and K2D3 is one of them) is unable to deconvolute the CD spectra or unwanted phenomena like aggregation, make this technique less applicable. In that case try other tools like FTIR, NMR, ESR etc.

The aim is not to know whether the alpha helical content is for example 68 or 69% but an overall picture and the way for example the secondary structure varies are the real features that might contribute to a further characterization and understanding of your (unknown) protein.

Hope this helps a bit for those picking up the challenge.

Best regards.