03-511/711, 15-495/856 Course Notes

03-511/711, 15-495/856 Course Notes - November 9, 2006

Introduction to Database Searching

BLOSUM Matrices

BLOSUM = BLOck SUbstitution Matrices
Trusted Data
- Much larger dataset
- Focus on conserved domains, MSA's are ungapped blocks.
Compute pairwise amino acid alignment counts
- Count amino acid replacement frequencies directly from columns in blocks (no trees.)
- Sample bias:
  - Cluster sequences that are x% similar.
  - Do not count amino acid pairs within a cluster.
  - Do count amino acid pairs across clusters, treating clusters as an "average sequence". Normalize by the number of sequences in the cluster.
BLOSUM x matrices
- Sequences that are x% similar were clustered during the construction of the matrix.
- Corrects for sample bias.
- Parameter of evolutionary divergence.

See Ewens and Grant, 6.5.2. for a detailed discussion of how the BLOSUM matrices are computed.

Comparing PAM and BLOSUM Matrices

	PAM	BLOSUM
Evolutionary model	Explicit evolutionary model	None
Data	Full length MSAs of closely related sequences. Assume no multiple substitutions.	Conserved blocks in protein families. Residues are under selective pressure
Bias correction	Trees	Clustering
Evolutionary distance	Obtained from Markov model of sequence evolution.	Obtained from clustering of sequences.
Matrices	Transition and log odds scoring matrices	Log odds scoring matrix only.
Parameter n	Distance increases with n	Distance decreases with n
Biophysical properties	Derived indirectly from data	Derived indirectly from data

The PAM and BLOSUM matrices were constructed from an evolutionary model and conserved blocks where amino acids are under selective constraints, respectively. Nevertheless, the matrices favor replacement of amino acids which share biochemical properties. Inspection of the BLOSUM 62 matrix shows that alignments of residues in the same biochemical group tend to have positive log odds scores. These residues are more likely to be observed together in related sequences than by chance. Residues from different groups tend to have negative scores. These residues are less likely to be observed together in related sequences than in chance alignments. A score of zero means that this pair of residues is equally likely in related and chance alignments.

Seq Identity	PAM	BLOSUM
20	250	45
30	160	62
40	120	80
50	80	-
60	60	-

Outline

Data base searching overview
- Applications
- Issues
  - Speed of search
  - Significance of matches found
Heuristics
- FASTA - not covered in this course.
- BLAST
  
  Today
  Original, ungapped BLAST: Altschul et al., '90
  Thursday
  Formal statistical model of ungapped local alignments
  Next Tuesday
  Gapped BLAST: Altschul et al. '97

Data base searching overview

Some applications:
- Compare cDNA with genomic DNA
  - To find gene location
  - To identify intron/exon boundaries
- Protein homology
  - Compare protein sequence with protein database
  - Structural and functional predictions
  - Evolutionary analyses
- Verification of gene prediction
Goals
- Find all high scoring local alignments (above some threshold)
- Determine significance of alignments found
  - Null hypothesis - sequences are unrelated
    p_j, p_k background frequencies
  - Alternate hypothesis - sequences are related
    q_jk target frequencies

Database searching is essentially a local alignment problem. In theory, a data base could be searched for sequences similar to a query sequence using dynamic programming. However, the running time for dynamic programing is O(mn), where m is the length of the query sequence and n is the length of the data base. For large data bases, the complexity of this approach is prohibitive.

The "typical" amino acid query sequence is 250 - 300 residues long, but query sequences could be much longer:

BRCA2
Genomic sequence:	86,101 bp
mRNA	10,257 bp
Protein	3,418 amino acids

Data base
11/8/2006
	Nucleic Acid Sequences	Amino Acid Sequences
Letters:	18,386,426,667	1,417,848,065
Sequences:	4,549,725	4,111,659

Last modified: November 9, 2006.
Maintained by Dannie Durand (durand@cs.cmu.edu) and Annette McLeod.