|
|||||||||||||||
 |
|
||||||||||||||
|
|||||||||||||||
| |||||||||||||||
|
Gibbs SamplingDescriptionGibbs sampling is a method that can be used to find weak signals or motifs in sequences, by the means of Monte Carlo sampling. Monte Carlo sampling is used to perform an effective sampling of the alignment space, by implemented two different moves - single sequence move - phase shift move In the single sequence move, a random place in the alignment is chosen and the sequence is randomly shifted in the phase shift move, the "alignment window" (See Figure 5) is randomly shifted, to escaping possible local minima. For accepting a move in the Monte Carlo sampling the energy of the alignment need to be low. The alignment energy is calculated based on a log-odds weight matrix as the log of the pseudocount and the sequence weight of amino acid a at position p divide by the background frequency af amino acid a.
The project is about reconstruction part of the work done in this article from CBS, namely
constructing a Gibbs sampler in perl for finding motifs of length 9 in protein sequences.
The paragraphs "Optimization of parameters for deriving amino acid sequences" and "Gibbs sampling" are especially important.
Input and output The output should be the aligned sequences. The motif should be apparent.
Examples of program execution:
Details |
