MEME Multiple EM for Motif Elicitation

1
MEME Multiple EM for Motif Elicitation

• ???
• 2005/08/19

2
Expectation Maximization

• Also call the Baum-Welch Algorithm
• The EM algorithm is useful in models and
  situations with hidden variables.
• Typical examples of hidden variable are missing
  or unobservable data, mixture parameter in a
  mixture models and hidden nodes states in
  graphical models (hidden units in NNs, hidden
  states in HMMs).

3
MEME

• MEME takes as input a group of DNA or protein
  sequences (the training set) and outputs as many
  motifs as requested.
• MEME uses statistical modeling techniques to
  automatically choose the best width, number of
  occurrences, and description for each motif.

4
Occurrences of motifs

• oops One Occurrence Per Sequence
• zoops Zero or One Occurrence Per Sequence
• anr Any Number of Repetitions

5
Input

• a group of DNA or protein sequences
• Fasta format

6
Output (1)
description of the sequences you submitted
7
Output (2)
Command summary (use default value if ignore )
8
Output (3)
degree of conservation at each motif
position
multilevel consensus sequence
9
Output (4)

• P-value and motif sequence

probability of getting a value of the test
statistic as extreme as or more extreme than that
observed
10
Output (4)

• BLOCKs format (for program extracting)

11
MEME command

• -dna sequences use DNA alphabet
• -protein sequences use protein alphabet
• -mod oopszoopsanr distribution of motifs
• -nmotifs ltnmotifsgt maximum number of motifs to
  find
• -minw ltminwgt minimum motif width
• -maxw ltmaxwgt maximum motif width
• -maxiter ltmaxitergt maximum EM iterations to run
• -evt ltevgt stop if motif E-value greater than
  ltevtgt

12
Practice

• Use pietty
• http//ntu.csie.org/piaip/pietty/stable/pietty032
  7.exe
• Host 140.115.156.80
• id bio1 bio11pw test1 test11
• ./meme.sh LTRxx.all
• Try to modify meme.sh