Current Topics In Genome Analysis Spring 2005 Week 4 .

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis ICurrent Topics in Genome AnalysisSpring 2005Week 4Biological Sequence Analysis IAndy Baxevanis, Ph.D.Overview Week 4: Comparative methods and concepts Similarity vs.vs. HomologyGlobal vs.vs. Local AlignmentsScoring MatricesBLASTBLAT Week 5: Predictive methods and concepts Profiles, patterns, motifs, and domains Secondary structure prediction Structures: VAST, Cn3D, and de novo prediction1

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IWhy do sequence alignments? Provide a measure of relatedness betweennucleotide or amino acid sequences Determining relatedness allows one to drawbiological inferences regarding structural relationshipsfunctional relationshipsevolutionary relationships importance of using correct terminologyDefining the Terms The quantitative measure: Similarity Always based on an observableUsually expressed as percent identityQuantify changes that occur as two sequences diverge substitutions insertions deletions Identify residues crucial for maintaining a protein’protein’sstructure or function High degrees of sequence similarity might imply a common evolutionary history possible commonality in biological function2

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IDefining the Terms The conclusion: Homology Genes are or are not homologous(not measured in degrees) Homology implies an evolutionary relationship The term “homolog”homolog” may apply to therelationship between genes separated by the event of speciation(orthology)orthology) between genes separated by the event of geneticduplication (paralogy(paralogy))Defining the Terms Orthologs Sequences are direct descendants of a sequence in acommon ancestor Most likely have similar domain structure, threedimensional structure, and biological function Paralogs Related through a gene duplication event Provides insight into “evolutionary innovation”innovation”(adapting a pre-existing gene product for a newfunction)3

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IDefining the TermsOrthologsA1B2Most recentcommon ancestorC3αDefining the TermsParalogsOrthologsMost recentcommon ancestorGene duplication A1B2C3A4αB5C6βGenes 1-3 are orthologousGenes 4-6 are orthologousAny pair of α and β genes are paralogous(genes related through a gene duplication event)4

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IOverview Week 4: Comparative methods and concepts Similarity vs.vs. HomologyGlobal vs.vs. Local AlignmentsScoring MatricesBLASTBLAT Week 5: Predictive methods and concepts Profiles, patterns, motifs, and domains Secondary structure prediction Structures: VAST, Cn3D, and de novo predictionGlobal Sequence Alignments Sequence comparison along the entire length ofthe two sequences being aligned Best for highly-similar sequences of similarlength As the degree of sequence similarity declines,global alignment methods tend to missimportant biological relationships5

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis ILocal Sequence Alignments Sequence comparison intended to find the mostsimilar regions in the two sequences beingaligned (“(“paired subsequences”subsequences”) Regions outside the area of local alignment areexcluded More than one local alignments could begenerated for any two sequences being compared Best for sequences that share some similarity, orfor sequences of different lengthsOverview Week 4: Comparative methods and concepts Similarity vs.vs. HomologyGlobal vs.vs. Local AlignmentsScoring MatricesBLASTBLAT Week 5: Predictive methods and concepts Profiles, patterns, motifs, and domains Secondary structure prediction Structures: VAST, Cn3D, and de novo prediction6

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IScoring Matrices Empirical weighting scheme to representbiology (side chain chemistry, structure, andfunction) Cys/Pro important for structure and function Trp has bulky side chain Lys/Arg have positively-charged side chainsScoring Matrices Conservation: What residues can substitute foranother residue and not adversely affect thefunction of the protein? Ile/Val - both small and hydrophobicSer/Thr - both polarConserve charge, size, hydrophobicity,hydrophobicity,other physicochemical factors Frequency: How often does a particularresidue occur amongst the entire constellationof proteins?7

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IScoring Matrices Importance of understanding scoring matrices Appear in all analyses involving sequencecomparison Implicitly represent particular evolutionary patterns Choice of matrix can strongly influence outcomesMatrix Structure: 1-1-1-1-1-1-1 Simple match/mismatch scoring scheme Assumes each nucleotide occurs 25% of the time8

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IMatrix Structure: 4-4-41BLOSUM62PAM Matrices Margaret Dayhoff and colleagues, 1978 Look at patterns of substitutions in highly relatedproteins ( 85% similar) within multiple sequencealignments Analysis documented 1572 changes in 71 groups ofproteins examined Substitution tables constructed based on results Given high degree of similarity within originalsequence set, results represent substitution patternthat would be expected over short evolutionarydistances9

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IPAM Matrices Short evolutionary distance change in function unlikely Point Accepted Mutation (PAM) The new side chain must function the same way asthe old one (“(“acceptance”acceptance”) On average, 1 PAM corresponds to 1 amino acidchange per 100 residues 1 PAM 1% divergence Extrapolate to predict patterns at longer evolutionarydistancesPAM Matrices: Assumptions All sites assumed to be equally mutable Replacement of amino acids is independent ofprevious mutations at the same position Replacement is independent of surroundingresidues Forces responsible for sequence evolution overshorter time spans are the same as those overlonger time spans10

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IPAM Matrices: Sources of Error Small, globular proteins of average compositionused to derive matrices Errors in PAM 1 are magnified up to PAM 250(only PAM 1 is based on direct observation) Does not account for conserved blocks or motifsBLOSUM Matrices Henikoff and Henikoff,Henikoff, 1992 BlocksBlocks SubstitutionSubstitution Matrix Look only for differences in conserved, ungappedregions of a protein family (“(“blocks”blocks”) Directly calculated, using no extrapolations More sensitive to detecting structural or functionalsubstitutions Generally perform better than PAM matrices forlocal similarity searches (Henikoff and Henikoff,Henikoff, 1993)11

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IBLOSUM n Calculated from sequences sharing no more than n%identity Contribution of sequences n% identical clustered andweighted to 1** * TEETSSQESAEEDKKPAQETEETSSQESAEEDA T Hook Domain (Block IPB000637B)2,000 blocks representing 500 groups of related proteinsBLOSUM n Clustering reduces contribution of closely-relatedsequences (less bias towards substitutions that occur inthe most closely related members of a family) Substitution frequencies are more heavily-influenced bysequences that are more divergent than this cutoff Reducing n yields more distantly-related sequences12

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis ISo many matrices.Triple-PAM Strategy (Altschul,Altschul, 1991)PAM 40PAM 160PAM 250Short alignments, highly similarDetecting known members of a protein familyLonger, weaker local alignments70-90%50-60% 30%BLOSUM (Henikoff,Henikoff, 1993)BLOSUM 90BLOSUM 80BLOSUM 62BLOSUM 30Short alignments, highly similarDetecting known members of a protein familyMost effective in finding all potential similaritiesLonger, weaker local alignments70-90%50-60%30-40% 30%So many matrices. Matrix EquivalenciesPAM 250PAM 160PAM 120 BLOSUM 45BLOSUM 62BLOSUM 80 Specialized matrices Transmembrane proteins Species-specific matricesWheeler, 200313

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis ISo many matrices.No single matrix isthe complete answer forall sequence comparisonsGaps Compensate for insertions and deletions Used to improve alignments between twosequences Must be kept to a reasonable number, to notreflect a biological implausible scenario( 1 gap per 20 residues good rule-of-thumb) Cannot be scored simply as a “match”match” or a“mismatch”mismatch”14

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IAffine Gap PenaltyFixed deduction for introducing a gap plusan additional deduction proportional to the length of the gapDeduction for a gap G LnwhereandGLn gap-opening penaltygap-extension penaltylength of the gapnuc pro51121Can adjust scores to make gap insertion more or lesspermissive, but most programs will use values of G and Lmost appropriate for the scoring matrix selectedOverview Week 4: Comparative methods and concepts Similarity vs.vs. HomologyGlobal vs.vs. Local AlignmentsScoring MatricesBLASTBLAT Week 5: Predictive methods and concepts Profiles, patterns, motifs, and domains Secondary structure prediction Structures: VAST, Cn3D, and de novo prediction15

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IBLAST Basic Local Alignment Search Tool Seeks high-scoring segment pairs (HSP) pair of sequences that can be aligned without gaps when aligned, have maximal aggregate score(score cannot be improved by extension or trimming) score must be above score threshhold S gapped or ungapped Results not limited to the “best HSP”HSP” for anygiven sequence pairBLAST AlgorithmsProgramQuery SequenceTarget teinBLASTXNucleotide,Proteinsix-frame translationTBLASTNProteinNucleotide,six-frame translationTBLASTXNucleotide,six-frame translationNucleotide,six-frame translation16

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis INeighborhood WordsQuery Word (W PQAPQNetc.1815141413131313131212 7 5 6NeighborhoodScore Threshold(T 13)High-Scoring Segment NQWIKQPLMDKNRIEERLNLVEA LA LTP G R W P D ER ATLASVLDCTVTPMGSRMLKRWLHMPVRDTRVLLERQQTIGA36533017

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis KQPLMDKNRIEERLNLVEA LA LTP G R W P D ER ve ScoreX365330Significance decay mismatches gap penaltiesSTExtensionScores and QWIKQPLMDKNRIEERLNLVEA LA LTP G R W P D ER rlin-Altschul EquationE kmNe-λSCumulative ScoreXmNmNλSk# letters in query# letters in databasesize of search spacenormalized scoreminor constantSTExtension18

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IScores and QWIKQPLMDKNRIEERLNLVEA LA LTP G R W P D ER ATLASVLDCTVTPMGSRMLKRWLHMPVRDTRVLLERQQTIGA365330E kmNe-λSCumulative ScoreXNumber of HSPsfound purely by chanceLower values signifyhigher similaritySTExtensionScores and QWIKQPLMDKNRIEERLNLVEA LA LTP G R W P D ER ve ScoreX365330E 10-6for nucleotidesE 10-3for proteinsSTExtension19

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis I20

NHGRI Current Topics in Genome Analysis 2005Biological Sequence Analysis IAvailable protein databases eference SequencesSWISS-PROTPatentsProtein Data BankLast 30 days21