Similarity Search

It has been shown that the TIM barrel domains of the two proteins flavocytochrome b2 (PDB code 1fcb) and phosphoribosylanthranilate isomerase (PDB code 1pii) are very similar in terms of their three-dimensional structure [Tang CL et al (2003) J Mol Biol 334(5):1043-1062; doi:10.1016/j.jmb.2003.10.025]. This similarity may not be detectable by traditional similarity searches.

1. Download yeast flavocytochrome b2 protein sequence in FASTA format from NCBI Entrez database (Hint: accession number NP_013658).
2. Perform a protein BLAST search http://blast.ncbi.nlm.nih.gov/Blast.cgi of flavocytochrome b2 against the PDB database using default parameters. (Summarize the number of hits, highest and lowest bit scores in a table).
3. Change the $E$-value to 0.01 and the word size from 3 to 2 and do the search again. (Do you see any differences in the number of hits? Can you find 1pii in the search result?)
4. Reset the $E$-value to 10. Change the substitution matrix from BLOSUM62 to BLOSUM45. (Compare the search results. What is your conclusion in terms of selectivity and sensitivity of your search?)
5. Reset the substitution matrix to BLOSUM62. Run the search with and without the low-complexity filter. (Record the number of hits, highest and lowest bit scores).
6. Run the same search as in item 2. using FASTA
   http://www.ebi.ac.uk/Tools/fasta33/ (Compare results with those from BLAST).
7. Return to the NCBI Blast homepage and repeat the search using the algorithms psi-blast and phi-blast. (Record the number of hits and try to find 1pii in the result page. Download all hits below $E$-value 0.005 in FASTA format).
8. Repeat the search with the hidden Markov model based approach HHPRED
   http://toolkit.tuebingen.mpg.de/hhpred. (Within the results can you find 1pii? Summarize how well different programs can find remote homologies.)
9. Goto the PDB database http://www.pdb.org/ and search for 1pii. Click on Tab ``3D Similarity" and check if you find some of the proteins in your previous similarity search results.