Sequence Alignment and Modeling System SAM

1
Sequence Alignment and Modeling System (SAM)

• UC, Stata Cruz

2
Quick Overview

• What is can do?
• Modeling, aligning and discriminating biological
  sequences (DNA, RNA, protein)
• Given
• A set of biological sequences (regard as
  related)
• SAM-T2K
• This method is available in SAM 3.0 or higher,
  used to build a linear HMM representing the given
  set

3
Linear HMM

• The model is a linear sequence of nodes,
  including match (square), insert (diamond) and
  delete (circle)

4
Applications and Contributions

• Protein structure prediction
• It is also an Automatic entries to CASP5
  (SAM-T99, SAM-T02) and CASP4 (SAM-T99)
• Yeast (S. cerervisiae) protein predictions
• SCOP super-family search and alignment library
  (not work now)
• Program update
• SAM-T98?SAM-T99?SAM-T02
• 22 publications

5
CASP

• Critical Assessment of techniques for Protein
  Structure Prediction
• Once 2 years
• CASP1 (1994)
• CASP2 (1996)
• CASP3 (1998)
• CASP4 (2000)
• CASP5 (2002)
• CASP6 (2004)
• CASP7 (2006), coming soon!

6
Program Structure (version 3.5)
Test set
Related sequences
Aligned sequences
Library of aa distribution
modelfromalign
(2)
Preliminary model
(3)
(1)
buildmodel
Refine model
HMM model
Evaluate
(1)
Fine-tuned HMM model
align2model
hmmscore
MSA results
Possible members
Protein DB
(2)
7
Primary programs

• Build a model
• buildmodel (from unaligned sequences)
• modelfromalign (from aligned sequences)
• Exeaming models
• makelogo and drawmodel
• Perform MSA
• align2model and prettyalign
• Scoring Sequences
• hmmscore
• Choose significant sequence entries by 2 scores
• target2k
• Use one sequence as seed to iteratively create
  model

8
Assignment

• Data
• Obtain a set of close related sequences, split
  into 2 parts training and testing
• Hint check known protein families
• Procedure
• Build a model for training set
• Evaluate the model by testing set
• Search all possible via HMM-based DB search
• Multiple alignment given and found
• Please read
• sam_doc.pdf (full manual)