Applied Genomic Technologies

Biological Databases

General Chemistry II

General Chemistry Lab

Organic Chemistry I

Organic Chemistry II

Organic Chemistry III

Organic Chemistry Laboratory I

Lecture 02  BLAST to Find Protein Function o If we have a protein sequence that we don’t know the function of, we can BLAST it to find similar proteins, and chances are that our protein does something similar o The output of BLAST is a series of protein sequences, aligned to the query sequence  The results are typically sorted in terms of similarity to the query, so the lower you go, the worse of a match you have  The alignments might not start at the same positions! o However, say we have two matches (match A and match B), where matches A and B have different functions  Naively, we might thing to assume our protein has the same function as match A  However, just because match A is the top-ranking hit doesn’t mean we necessarily should assume it’s the correct functional match!  The Computational Question o How do we measure similarity between sequences? Percent identity? o Given two sequences, how do we compute a “good” alignment?  Computing Alignments o Make a 2 by m table, where m is the length of the alignment o Each of the two rows is a sequence, with interspersed gaps o Columns describe the edit operations o Distance : Some cost related to the number of changes to go from sequence 1 to another sequence o Each column gives us the operation we used (keep, delete, change, insert), and we can assign each operation a cost  The score of the alignment would be the sum of the score of every column o How do we compute the alignment with the best score?  A First Guess o We could try all alignments and choose the alignment with the best score o However, this would take ( m m − n ) = ( m n ) = 2 n ∗ 2 m − n ≅ 2 m o This is VERY time-expensive!  Optimum Scoring Alignments and Score of Optimum Alignment o Instead of computing the actual alignment itself, we will actually compute just the optimum score o Observations : The optimum alignment has nice recursive properties  The alignment score is the sum of the scores of the columns  If we break off at cell k , the left part and right part must be optimal sub- alignments (so we can draw a line anywhere and both parts will be optimal)  The left part contains prefixes s[1… i ] and t[1… j ] for some i and j (we don’t know values of i and j ) o We might think to use Divide-and-Conquer, but it turns out that it’ll also be very time-expensive to implement o The solution is to use Dynamic Programming :  Compute the best alignment of a prefix containing i characters in the first sequence ( s ) and j characters in the second sequence ( t )  Let’s say the last column of the alignment is column k  Say we have our “score function” S ( i,j ) that gives the optimum score from s 1 to s i and t 1 to t j  Also say we have our “column score function” C ( s i ,t j ) that gives the column score for a given column  We have three possible cases:  [ s i t j ] (a match) o Everything to the left of k is an optimum alignment of s[1… i -1] and t[1… j -1] o S ( i,j ) = S ( i − 1 ,j − 1 ) + C ( s i ,t j )  ¿ (a gap for sequence t ) o Everything to the left of k is an optimum alignment of s[1… i -1] and t[1… j ] o S ( i,j ) = S ( i − 1 ,j ) + C ¿  [ − ¿ t j ] (a gap for sequence s ) o Everything to the left of k is an optimum alignment of s[1… i ] and t[1… j -1] o S ( i,j ) = S ( i,j − 1 ) + C ( − ,t j ) o Thus, the solution:

CHEM | 182 | Biological Databases | lecture notes | Lecture 02  BLAST to Find Pro

Lecture 08  Finding TF Motifs o De Novo or Ab Initio : Finding TF binding sites without prior information o You could take the promoter sequences and do a local alignment, and a region that is highly conserved is a good place to look for the motif o We could also use ChIP-Seq  Protein Recognition o We can use antibodies to detect proteins o Antibody : A protein that binds to a specific antigen  Can bind to small molecules, or even other proteins o Wide range of applications  Drug testing, assessment/monitoring of cancers, detection of specific metabolites, identification of pathogens, monitoring infectious agents o Limitations:  For proteins, the gene must be expressed  Antibodies o Polyclonal Antibodies : Antibodies that come from different B-cells (so they might target different parts of the antigen, less specific) o Monoclonal Antibodies : Antibodies that come from clones of the same B-cell (so they all target the same part of the antigen, more specific)  Since adult B-cells don’t clone on their own, we have to fuse it with a tumor cells o So we have B-cells, tumor cells, and fused cells, how do we isolate fused cells?  B-cells will die out on their own, so they’re fine  What we can do is have some deficiency in the tumor cells that exists in the B-cells, and provide supplements to keep them alive for a bit  Then, once we think stuff has fused, we can stop giving the supplements  The B-cells will die on their own  The tumor cells will be starved and die  The only cells left will be the fused cells

CHEM | 183 | Applied Genomic Technologies | lecture notes | Lecture 08  Finding TF Motifs

ps6keyfinal2.pdf

Problem Set #4, 5.12 Spring 2003 Due Monday, March 10, 4pm 1. a) Label each pair as enantiomers , diastereomers , or same molecule. b) Label each stereocenter with its R or S configuration. Br and F D H F D Br H OH O Br HO CH 3 and HO CH 3 O Br OH Cl Cl Cl and Cl Ph H H OH CH 3 and H 3 C H 3 C Ph OH CH 3 2. a) Label each molecule as chiral or achiral . b) Label each stereocenter with its R or

prob_04.pdf-Problem Set #4, 5.12 Spring 2003

exam2_s05.pdf

All courses in CHEM

Popular Documents