Dear Researchers
To find the sequence entropy of a protein (say, pdb code 4ICB) / DNA , I am using some codes written in C program. For this, I need to use BLAST and ClustalW2 , and resulting aligned sequences are fed into the program. I am curious to know whether there is any handy way of finding sequence entropy of a protein / DNA.
Thank you for help,
Mahendra Thapa
Hello, Mahendra!
What do you mean when you say the "sequence entropy"?
Entropy is the logarithm of the phase space volume for the particular model/system.
So, to estimate the entropy, you have to define the generalized coordinates of the problem under study, then estimate the volume of thus defined phase space, then take the log of it. That simple :-)
It is not so often done, to be honest. To give you a hint, you can take a look at this arxiv paper, pages 3 and 5: http://arxiv.org/pdf/1503.06840v1.pdf
I continue working in that direction, though it is not straightforward to get entropy estimates for protein folding problem.
I think what he's looking for is sequence entropy (as defined by Shannon) instead of thermodynamical entropy (which is indeed much harder to calculate). In that case, it is pretty straightforward - just calculate the frequency of each of the 20 residues on each positions (some people suggest using the gap as a 21st residue), then the entropy for each position will be -sum(1 to 20)P log P. (for a properly formatted equation, see this site for example: http://imed.med.ucm.es/PVS/pvs-help.html ).
Yeah he said that clearly -- 'sequence entropy', my fault. Sorry for misunderstanding.
Thank you very much Dr. Artem and Dr.Lucas for the suggestions.
Bio3D program developed by 'The Grant Lab' also gives the 'sequence entropy'. I was just looking for alternative & handy ways.
Hi,
Maybe this tool from the HIV database will work for you: http://www.hiv.lanl.gov/content/sequence/ENTROPY/entropy_one.html
Well... any well-written computer program does exactly what the programmer has implemented. Programmer, when writing the code, has a MODEL in his/her head. Depending on the MODEL ASSUMPTIONS and the APPROACH, results are different.
So, when you say, took the output from one code, which served as input for another, you clearly and explicitly confess, that you have no glue, what are you doing, in fact.
Now. There is no a thing, called "sequence entropy", in Statistical Mechanics textbooks. If you introduce a new terminology, please define it. At some point you will need to make assumptions regarding averaging over sequences, which is not a trivial question. Once you formulate your question well enough, you will get a half of answer. With the other half we might be able to help.
I have my own implementation for what you need. Requires a sequence (any) and provides the value of the Shannon's entropy:
https://github.com/Gagniuc/Entropy-of-Text
However, shannon's entropy doesn't take into account the order of the symbols. Thus, I have developed a new method, a new algorithm for measuring information, which takes into account this order of the symbols. The algorithm is called self sequence alignment, please see:
https://github.com/Gagniuc/Self-sequence-alignment
Also see the pdf describing the method here:
Presentation Self-Sequence Alignment - Implementation vs model
A comparison between shannon entropy and my method:
https://github.com/Gagniuc/Information-Content-vs-Information-Entropy
- Badges
- Science topic
- Similar topics
- Engineering
- Chemical Engineering
- Membranes