Practical 5: Homology searching

This practical is intended to give you a feeling for the programs and data structures encountered in database searching. Most of the frequently used programs come with online tutorials, for example BLAST tutorial. You are encouraged to experiment with different search programs and parameter settings. Look carfully at the results and compare it with what you expected or wanted to achieve. Below you will find four different types of problem that you will need to work on. If you would have to wait for a server because of user load, you can start working on another of the four examples in the meantime, so to be able to finish the assignments within the allocated time today. PLEASE NOTE: For this assignment it is crucial that you clearly write down the procedure you followed, which results you observed from your 'searches' etc. Simple yes/no answering will not be appreciated. Tip: Open links with right mouse click -> 'Open Link in New Window'. Example 1: Insulin Insulin is a protein hormone that is secreted into the blood by the pancreas. The function of insulin is to regulate the blood glucose level (together with other regulatory proteins such as glucagon). Vertebrates need a very constant blood sugar level to regulate the body temperature and to supply the body with energy (e.g. brain, muscles). Insulin has evolutionary developed from a family of protein growth factors, the Insulin-like growth factors. Access the Entrez server at NCBI. Start a -Search- 'Protein' -for- 'insulin'. Try also the keyword 'insulin precursor human'. The sequence we want to retrieve is P01308 [gi:124617]. Find the protein sequence in the original SwissProt database. We want to get it in FASTA format. The FASTA format of a sequence contains a single header line starting with a '>' character and a description of the sequence origin and status. The lines below the header line contain the protein sequence in single-letter code for amino acids or nucleotides (depending on the type of database). Details about homology, function or structure and similar information in the FASTA header are called 'annotation'; they are an important source of information. The sequence should look like the example insulin sequence. Q1. Provide the FASTA sequence We perform an iterative sequence databank search using PSI-BLAST on the non-redundant Genbank database. Choose 'protein blast' at the BLAST site. Next, under 'Program Selection', select PSI-BLAST. Paste the insulin sequence in FASTA format into the text window for the query, choose 'nr' for 'non-redundant' database, and execute the search. Look through the best hits: there are insulin, proinsulin (before proteolytic cleavage of connecting C-peptide), and pre-proinsulin (before proteolytic cleavage of signal peptide). Some entries are single chains (A or B chain) of insulin. Q2a. What is your best hit? Is this what you expected? Q2b. What is the meaning of the bit-score and the corresponding E-value? Q2c. What is the correlation between these two? Try to find the distantly related 'insulin-like growth factors' in the result list. Q3a. How many of these did you retrieve? Now run a second PSI-Blast iteration. Q3b. Did the number of 'insulin-like growth factors' change? Repeat the search choosing 'BLOSUM45' as matrix, set 'Expect threshold=100' and 'Max target sequences=5000' (you will find these options under 'Algorithm parameters'). BLOSUM45 (=>45% identity) is derived from more distant sequences than the de fault BLOSUM62. The insulin-like growth factors have much lower (better) E-values than in the first search, because the BLOSUM45 matrix favours more distant sequences (i.e., the matrix is not as hard as the BLOSUM62 matrix). Q4. Do you see some sequences from distant organisms (for example insects)? Also write down the accession numbers. For comparison use the ScanPS (installed at the European Bioinformatics Institute) with the same sequence. Read the instructions before sending the search. Note the difference: this search is based on sequence comparison by dynamic programming, whereas BLAST uses word match and extension, with a final dynamic programming step for promising hits only. Try the PAM250 matrix (increasing numbers in PAM - Percent Accepted Mutation - indicate increasing sequence distance. PAM250 means 250 mutation events have occurred per 100 residues.) Q5. Do you find more distant sequences? Example 2: EF-hand motif The EF-hand is a helix-loop-helix structure with a specific sequence and configuration of complexing residue sidechains in loop positions 1,3,5,7,9 and 12. The specific sequence is called EF-hand motif. The EF-hand binds Ca2+ ions with high affinity, and EF-hands are generally paired. A well studied protein carrying EF-hands is calmodulin, which undergoes a dramatic conformational change upon binding of Ca2+ ions. This conformational transition from the Ca2+ free (apo) form to the Ca2+ bound (holo) form exposes hydrophobic surface area and enables calmodulin to bind to other proteins. This mechanism allows calmodulin to mediate changes in cellular Ca2+ levels by binding to other proteins. For example, a complex is formed with myosin upon Ca2+ influx in the process of muscle contraction. Consult the library of EF-hands for more details. Extract the calmodulin entry (P62146) from the SwissProt database. Find the sequence of the EF-hands (from the 'Feature table') and save them as a multiple alignment (stacked sequences). Q6. Provide the alignment. Does it look like the following example: EF-hand sequences? Note that these sequences are so short that sequence searching will yield only near-identical hits (if at all), because each matching character needs to score high in order to produce a total score above the threshold. This problem can be circumvented by motif search. Motifs can be described by string patterns. A widely-used database of functional motifs is the ProSite database. This database also harbours software to construct a motif from a multiple alignment and software to scan the SwissProt sequence database with a given motif or to scan the ProSite motif database with a given protein sequence. Look at lecture slides 55-58 to get a feel of the ProSite database and associated software. Become familiar with the pattern syntax (look for section "Settings - Pattern syntax" on the page). Translate the multiple alignment into a ProSite motif. Q7. Provide the ProSite motif. Paste the motif into the query field (top right) at ProSite. Your search pattern could look like the example pattern, but try also variations to see the effect on the search result. Q8. How many EF-hands do you find? REMEMBER: If you do not know, understand or have any clue what you are doing, just ask. That's why we are there!

(c) IBIVU 2026. If you are experiencing problems with the site, please contact the webmaster.