em

Function

Represents the EM algorithm as used by MEME.

em(profile,dataset_start,t,l,oops_model)

em(profile,dataset_start,t,l,gamma,zoops_model)

em(profile,dataset_start,t,l,lambda,tcm_model)

Include Headers

seqan/find_motif.h

Parameters

profile	A StringSet of frequency distributions. Types: StringSet
dataset_start	An iterator pointing to the first input sequence of a given dataset. Types: Iterator
t	The number of input sequences.
l	The size of the motif.
oops_model	The oops_model object. Types: Oops
zoops_model	The zoops_model object. Types: Zoops
tcm_model	The tcm_model object. Types: Tcm
gamma	The probability of sequence having a motif occurence.
lambda	The probability of starting a motif occurence Remarks: lambda is calculated by dividing gamma by the length of the corresponding sequence.

Remarks

This version of EM is used in the MEME program of Bailey and Elkan. It is a Bayesian variant of the basic EM which allows multiple occurrences of a motif in any sequence and can therefore be performed on sequences of one of the model types Oops, Zoops and Tcm. We use the EM algorithm of MEME for the refinement step of PROJECTION.

SeqAn - Sequence Analysis Library - www.seqan.de