Flashcards
BIOL 365 midterm2
Cards in this deck (170)
- 1Define bioinformatics, why it was adopted and how it was born.research, developmt, appliction of comp tools to compile and analyze data
adopted due to increasing sequence data - 2What’s the difference between clustering & phylogenetics?Clustering = grouping by similarity; changes can be additive/not
Phylogenetics = clustering nt/protein seq by similarity, detailed, ADDITIVE & FOLLOW EVOLUTION MODEL - 3Describe characteristics of a flat file database. Adv?Data stored as txt file (and collections)
adv: easy to input & distribute - 4Describe characteristics of a relational database. Adv?data stored in tables linked by shared field (key) unique to record
adv: handles more data, decrease data memory, faster search + retrieval - 5What's the relationship between primary and secondary data? Why is functional annotation useful?Primary: from experimental results
Secondary: derived from primary data
Fxnal annotation can help det. data meaningfulness - 6What's the main goal of data QC? How is it done (mostly)?Goal: decrease redundant seq.
Done via automation (manual curation higher quality) - 7Which 3 databases make up the international nt seq databse collab (INSDC)?Genbank (NCBI)
ENA
DNA Data Bank of Japan (DDBJ) - 8What's the relationship between Swiss-Prot & TrEMBL? Where are they accessed from?Swiss-Prot = curated protein db
TreEMBL = comp annotated supplemt to Swiss-Prot
Both accessed via ExPASy - 9What is Ensembl? (I.e. type of data it stores)Annotated genome browsers (stores genome, duh)
- 10Which database is used to find protein structure?PDB
- 11What's included in genomic & gene data for NCBI?chromosomal DNA, large DNA frag, full gene, seq tagged sites (STS)
- 12What's included in cDNA and EST?cDNA, EST (partial copy of RNA transcript)
- 13What can be found via Protein database of NCBI?seq'd protein
trnslted mRNA (via DNA)
seq to known protein structure - 14What are the 2 types of data under "curated seq data"?RefSeq
UniGene - 15What's the goal of RefSeq? How is it tailored?unify seq record to provide most reliable seq data: gives best rep. ref. seq (i.e. 1 refseq/gene/organism)
- 16What's the goal of UniGene and how is it made? What can it be used for?give 1 entry for gene & collect EST associated with gene via clustering similar EST
Used to det. expression pattern - 17What's the purpose of an accession #?Unique ID for specific seqs
- 18What type of info/links does PubMed provide?Link to biomed literature
- 19What are defined by controlled vocabularies and ontologies?CV = define data type
Ontology = define data relation - 20What is MeSH and what are some of its traits & fxns?Medical Subject Heading
ctrled vocab indexing of pubmed articles & used to refine search
CURATED & ctsly updated - 21What is Taxbrowser used for and some of the links/infos it provides?search lineage + scientific name of organism; provide link to other resources
Entrez for nt & protein seq, PopSet - 22Is OMIM a part of NCBI? What does the database document & info it might return?Online Mendelian Inheritance of Man
NOT IN NCBI
catalog of human genes/genetic disease
link to PubMed, Entrez, gene maps - 23What are 4 purposes of seq alignment?1. det if 2 seq evolutionarily related
2. ID common domains & motifs
3. ID conserved regions vs. variability
4. genome annotation* - 24Define seq alignment.pw match between chars in each sequence while preserving order
- 25Why are SCORES and GAPS essential to optimal sequence alignment?Scores = define criteria for algorithm to choose best alignment
Gap = intro'd to max overall score - 26Define homologs; unique trait?Seqs that share a common ancestor (related)
YES/NO question - 27Differentiate between orthologs, paralogs, xenologs + ex. For each.Ortholog - related seq in diff SPECIES that diverged via speciation (ex. whale vs. horse Mb)
Paralog - related seq in genome sharing similarity via duplicated ancestral gene (ex. human alpha/beta chain of Hb)
Xenolog - lateral gene transfer relationship (ex. plasmids) - 28Differentiate between the concept of % identity vs. Similarity.% ID = % of exact matches in aligned seq
similarity = approximate matches (ex. aa fxns) - 29Define pairwise alignment. Why might 2 nonhomologous seq appear highly similar?comparison of 2 seq & looking for matching chars.
nonhomolog similar due to convergent evol., parallel evol., reversion - 30What are the advantages of nucleotide vs. Protein alignment? Ideal approach?Nt: confirm cDNA ID, study noncoding DNA, study polymorphism, comparison within species
proteins: more informative, codon degeneracy in 3rd position dosn't affect aa, less noise! - 31Differentiate between global vs. Local alignment (brief). Give examples.Global = align entire seq end to end (N&W)
Local = seq stretch w. highest density of matches aligned (S&W) - 32What are global & local alignment respectively useful for?Global: comparison w/in species
Local: multidomain proteins - 33Name 3 methods under pair-wise seq alignment.1. Dot plot
2. Dynamic Programming
3. Heuristic methods - 34Explain how dot plots work.2 seq in 2D array
dots placed in boxes where matches occur
regions of identity = diagonal line - 35Explain how sliding windows work & adv.detect matching region via filtering with sliding window that filter out random match
specifies window size & stringency to capture regions of SIG similarity - 36What is the rationale for gap penalties? What would be a better alignment?Help get best alignment, but mutation/insertion is rare
Best = fewer but extended gaps - 37How does gap penalty formula justify indels? (give eqn)Gap penalty = G+nL
Favors seq with few long indels instead of MANY small ones - 38Give definition and the functions of dynamic programming algorithm.Break down big problem to smaller subproblems & iteratively use previous solution to get to the next.
Max. matches of chars - 39Explain how dynamic programming algorithms work. What’s a major advantage?Compares every pair of char. for 2 seqs & generate alignment by breaking big problems to subproblems + saving their solutions in table
Adv = guaranteed optimal - 40Briefly describe how scoring works for Needleman-Wunsch algorithm. What type of alignment is this?2D array is used to compare 2 seqs & store value for partial seq alignment.
Each entry = best score for alignment until that point based on gap, mismatch, match
Global alignmt - 41For what instances are local alignments useful? Given an example.Recog small sub-seq in coding region
Several regions in diff order in complete seq
ex. multidomain protein, DNA region w. several genes - 42Why can’t global alignments work for multidomain proteins?Optimal score depends on DOMAIN ORDER (might not be true in multidomain protein w. few matches)
- 43Briefly describes how the Smith-Waterman algorithm works. (hint: 2 new rules)Same 2D array sys as N&W but w. more severe mismatch penalty
new rules: 1) zero is new min, 2) traceback from max. value instead of bottom right & stop when reaching zero - 44Name the advantages and disadvantages of global alignment.Adv: highest total score; useful for seq with similar L
Disadv: regions of high local similarity ignored; depend on order of genes/domains - 45Name the advantages and disadvantages of local alignment.Adv: ID local similarity that doesn't span entire L; sensitive
Disadv: higher rate of false +, computationally expensive - 46Briefly describe the general approach to PW alignment (5 points)1. choose 2 seq
2. pick algorithm gives score (global vs. local)
3. Set score for mismatch, match, gaps
4. asses total score for similarity
5. estimate probability that alignmt occured via chance (E-value) - 47What are the 2 basis for aa scoring functions?1. purely statistical approach (i.e. freq of exchange)
2. fxn of biochemical properties (specific to aa) - 48What is the BLASTN matrix score?For nucleotides based on matrix match/mismatch scores + gap penalty
- 49What are the possible consequences from changes in nt sequence (3)?1. change in aa
2. synonymous sub
3. chain termination/elongation - 50What is “fxnal constraint” wrt to aa substitution.AA may be req'd for fxn/protein structure > will have lower observed rate of change
- 51What does the observed probability of an aa exchange depend on? (4 factors)1. fxnal constraint
2. natural aa freq.
3. codon freq & bias
4. mutability of aa (fxnal stability + chemical similarity) - 52What makes up a substitution matrix & how is it constructed? What should it ultimately show?Values based on probability that aa1 mutates aaj for all 20 pairs
Constructed via assembling large & diverse verified MSA/pw for aa.
Reflect avg probability for each possible aa change. - 53How does pairwise alignment of protein sequences differ from nt pairwise alignment?Scoring matrices derived from actual substitution in nature; each aa has specific scores
- 54What is the scoring matrix based upon & what 2 models are used for comparison?Based on probabilistic approach + model then used to generate the odds ratio.
2 models: 1) evolutionary 2) random - 55What’s the odds ratio?Probability alignmt authentic/probability alignmt random
- 56What assumption is used in the derivation of scoring matrices? How to get total odds score?Assume each site is INDEPENDENT
Total odds = product of odd ratio for each column - 57Why is the log ratio used and how is it interpreted? (i.e. meaning for a +/- value).Used because can sum up total
+ = 2aa substitution more likely
- = 2aa substitution less likely - 58What is the PAM matrix & what is it based upon? What does it reward & penalize?Based on global alignment of CLOSELY related proteins; evolution basis that aa change need to be accepted by species as dominant form
Reward match; penalize mistmatch - 59What is Mij? How is this interpreted (ex. PAM1)?Mutation probability matrix
Probability i is replaced by j over a given interval (% or log10)
Interpreted as unit of divergence as # of change/100 aa - 60How does the degree of divergence differ wrt diff. PAM? (hint: diff/100 residues)Lower PAM = less divergent (fewer changes)
Higher PAM = more divergent (more change) - 61What is the “twilight zone”?Asymptotic region of PAM matrices where matches that indicate homology becomes questionable
- 62How are BLOSUM matrices derived + what it stands for?Blocks Substitution Matrices
Based on BLOCKS database with ungapped MSA for local alignment of distantly related proteins - 63Name 3 differences of BLOSUM relative to PAM matrices.1. based on distantly related proteins
2. seq directly compared (no evolutionary basis)
3. goal is to ID conserved regions in seq - 64What is done to prevent bias in BLOSUM matrices?Via similarity thresholds that collapse seq. above this into one cluster
- 65Which PAM and BLOSUM would be chosen for more divergent proteins? Less divergent?More divergent: higher PAM, lower BLOSUM
Less divergent: lower PAM, higher BLOSUM - 66What approach is used to account for gaps?Gap penalties
affine vs. non-affine (linear) - 67What approach is used to account for gaps?Gap penalties that include origination + extension
- 68What is meant by sensitivity wrt a search algorithm?ability to find true + & avoid false -
- 69Briefly describe what is BLAST.Basic Local Alignment Search Tool
Set of algorithm that compares query to all seq in a data based via pw comparison to reflect similarity. - 70Briefly describe the BLAST algorithm.1. Make list of word pairs above threshold T
2. Search for neighborhood word hit > T
3. From hit word, make outward extension for exact word match = seed for ungapped alignment
4. tidy up, calc. indels & match score - 71Name 4 applications that BLAST can be used for.1. ID ortholog & paralog*
2. discovery new genes/proteins
3. discover gene/protein variants
4. investigate expressed seq tags (ESTs) - 72Describe the functions of BLASTP, BLASTN, BLASTX, TBLASTN and TBLASTX.blastp - protein/protein query
blastn - nt/nt query
blastx - dna trslted 6X to query protein
tblastn - protein queried against dna trslted 6X to protein
tblastx - query+database protein trnslted 6X to make 36 queries - 73When might protein sequence alignment be more useful?for inferring structure/fxn from seq
- 74Why should low complexity regions be filtered out?Will give high similarity/% ID without being related to query (false +)
- 75What is the relationship between similarity & homology?Significant similarity usually indicates homology but homology not always give seq similarity
- 76What effects do words and threshold have on alignment results?large w = high sensitivity, slower search
small w = lower sensitivity, faster search
higher T = faster search
lower T = slower search - 77What type of distribution do local alignments & BLAST results follow? Why?extreme value distribution (EVD)
Search results give high scoring pairs (HSPs) that skew the distribution closer to upper end - 78Define the expect value E.# of alignments w. score >= to score S that are expected to occur by chance during a search
- 79How is E value related to probability value P? What happens when E is really small?E value related to p but easier to interpret.
When E is really small, E~p (<0.05) - 80Write the function for expect value; what does each variable represent?E = Kmn e^(-lamdaS)
K = scaling term
m = length of query
n = database size
lambda = fxn
S = score - 81Why must S for aligning a pair of random sequence on average be negativeOtherwise long random alignments will get very small E-vaues (false significance)
- 82What does it mean when E=1?One match w. similar score is expected to occur by chance within a random distribution
- 83What are the differences between raw vs. bit scores?Raw score = calc'd from substitution matrix
Bit score = normalized - 84What are the advantages of bit scores?allow result comparison from diff. databases (even if scoring matrices are diff)
- 85What’s defined as a true positive?True homolog seq to query
- 86How can homology be identified (2 factors)? What other info might be useful?1. seq similarity
2. E-value
Can also use biol info & multiple comparisons - 87Name 5 ways to decrease # of results?1. Lower E-value
2. Restrict organism
3. search seq subset
4. change scoring matrix
5. restrict to RefSeq - 88Name 5 ways to increase # of results?1. Increase E value
2. Restrict search space to organism
3. change scoring matrix to more divergent (lower BLOSUM)
4. remove database restrictions (decrease word size?)
5. Sensitive approach (PSI-BLAST, MSA) - 89What is PSSM and how does it differ from BLOSUM and PAM? Adv?Position-specific matrix
Made via MSA of seq homologous to input query & integrate residue conservation
Adv = customized; capture patterns of conservation + substitution - 90Briefly, what’s the key purpose of PSSM?To dig deeper into the database & ID remote homology
- 91Describe the 5 steps used to perform PSI-BLAST.1. search query against database
2. make MSA to create profile (specialised PSSM)
3. PSSM used as query against database
4. PSI-BLAST estimate E-value
5. Repeat 3&4 to iteratively refine search - 92What’s the structure of the PSSM?Dimension L seq x 20aas
Freq of each aa at each position of alignment weighted & contribute to score - 93How would conserved vs. nonconserved region be identified on a PSSM?Conserved = + value surrounded by lots of very - values
Nonconserved = values are all approx. 0 - 94How is a PSSM constructed?1. MSA used as input to build profile
2. Use log odds method to calc. diff score for each position in profile (observed freq + pseudocount freq in BLOSUM) - 95What is the advantage and disadvantage of PSI-BLAST?Adv: deeper digging of database, more inclusive of remote homologies
Disadv: corruption (once included, won't leave for the rest of iterations) - 96Formally define corruption wrt PSI-BLAST. What are 3 approaches to stopping it?Presence of at least 1 false + with E < 10^-4 after 5 iterations
Stop: 1) filter biased regions, 2) adjust to lower E, 3) remove suspicious hits via visual inspection - 97What is RPS-BLAST?Searches pre-existing PSSM that best suit input & queried against BLAST
NO ITERATION - 98Briefly describe how RPS-BLAST works.1. Input query
2. query compared to existing PSSM database
3. BLAST-like procedure used to align query to existing PSSM & generate score
4. RPS-BLAST estimate significance (E-value) - 99How does Delta-BLAST work?ID existing PSSM in first iteration and use it to search protein database
- 100Define multiple sequence alignment & 2 ways residues can be homologousalign 3 or more seqs
1) evolutionary homolog.
2) structural homolog. - 101What are some properties of MSA?Don't have 1 correct alignmt; may not be possible to ID aa that align in structural sense in MSA
- 102What are 4 features of AA that help to characterize MSA results?1. known structural motifs
2. conserved motifs
3. secondary structure features
4. regions with pattern of indels - 103What are advantages of MSA relative to pairwise alignment? (5)1. more sensitive
2. reveal conserved residue/motif
3. analyze population data
4. single query can be searched against MSA database
5. ID consensus seq - 104What are the 3 factors to consider when including seq in MSA?1. # of seq to include (more!)
2. remove redundant seq
3. remove questionable seq - 105What is Clustal-W and how does it work (steps)?Progressive alignment that allow seqs to be globally aligned in specified order
1. do set of pw global alignment
2. create guide tree w. order
3. group seq together from the most similar until all are included in MSA - 106What is an advantage and disadvantage of Clustal-W?Adv: rapid alignmt of many seq
Disadv: final alignment depend on seq order (inaccurate) - 107What are the differences between guide vs. phylogenetic trees?Guide trees based on pw similarity, not MSA similarity (intermediate step).
Phylogenetic tree uses MSA similarity to demonstrate ancestor-descendant relationships - 108What are two kinds of multiple sequence alignment resources (give examples)?1. Database of MSA (ex. BLAST)
2. MSA via manual input (ex. ClustalW, MUSCLE) - 109How to reduce amplification of initial error in progressive MSA given this is a greedy algorithm?1. weight pw alignmt via other seq
2. iteratively remove & realign seq. - 110How do iteratively re-alignments generally work?Remove subseq & realign using tree-based hierarchy
- 111Briefly describe how MAFFT works.1. divide progressive alignmt via tree-dependent partitioning
2. realign partition then realign subgroup
3. if objective score better, replace initial & repeat - 112What does muscle stand for and what are the 3 general stages?MUltiple Sequece Comparison using Log-Expectation
1) Draft progressive
2) Improved progressive
3) Refinemt - 113Describe how the MUSCLE algorithm works.1. estimate guide tree #1 via kmer counting
2. PA to make MSA#1
3. compute % ID -> generate new distance matrix
4. make guide tree #2 from new DM
5. build MSA #2
6. dlt each edge of tree #2 via decreasing dist.
7. compute profile of 2 subtree
8. realign profile from #7 = new MSA; keep if better
9. repeat #6-8 until reaches convergence - 114How is each individual sequence realigned to rest of MSA in MUSCLE?Edges are dlted in order of decreasing distance from tree root; subtree also realigned via guide tree order
- 115What are some advantages of MUSCLE?1.Seq realigned to rest of MSA
2. fast & good accuracy wrt other heuristic method - 116What is phylogeny defined as & how is it constructed?Hypothesis of ancestor-descendent relationships
Via molecular seq. data - 117Describe the molecular clock hypothesis & how is it derived?For given protein, rate of molecular evolution is approximately constant in all evolutionary lineage.
Plot x-axis = divergence times of species, y-axis as corrected # of aa change/100 residues - 118What are 3 main conclusions derived from plot of molecular clock ?1. rate of aa substitution constant for each protein lineage
2. avg rate of change DIFFER betwn diff 2 lineage
3. observed variation in rate of change = fxnal constraints via natural selection - 119Describe the neutral theory of molecular evolution.Kimura
Most DNA change not selected via fitness; main reason is random drift of mutant alleles that are selectively neutral - 120What’s balancing selection?Selection done to maintain genetic variation
- 121How is molecular phylogeny demonstrated in bioinformatics? (ex. PAM & Clustal)PAM - substitution score based on odds of mutation occurrence AND organism acceptance
Clustal - seq assumed a priori to share common ancestor - 122Define topology, taxon, OTU, internal nodes and root.topology - tree branching pattern
taxon - distinct gp of organism (node)
OTU - taxa under consideration (external node)
internal nodes - hypothetical ancestor
root - common ancestor for all OTU studied - 123What’s the difference between a cladogram vs a phylogram?Cladogram - unscaled branch; OTUs equidistant from root
Phylogram - scaled branch length proportional to evol. distance - 124What is the purpose of tree root and how can a tree be rooted?Root = common ancestor of seq & provide tree directionality
Rooted via outgroup w. distantly related taxon from all other gps - 125What are the traits of a Newick format?Topology & distance recorded as nested brackets specifying hierarchy
- 126What’s the difference between an ultrametric tree vs. an additive tree?Ultrametric - assume CONSTANT mutation rate; dist. root to tip equal for all taxon
Additive - rooted via an outgroup - 127How does info inferred from species tree differ from gene/protein tree?Species - each internal node is speciation event (true orthologs)
Gene/protein tree - internal nodes not necessarily speciation (can be due to gene duplication/loss)
SPECIES RELATIONSHIP IS NOT GENE RELATIONSHIP!!! - 128Why should phylogenetic trees be additive?Preserve sum of distance betwn OTUs within in tree; mutation is also cumulative
- 129What is UPGMA and how does it work?Unweighted Pair Group Method with Arithmetic Mean
Combine 2 closests gps at each iteration to form new gp & continued until make rooted tree - 130Briefly describe the steps used to generate UPGMA?1. calc pw distance betwn taxa (ex. p distance)
2. combine closest taxa pair
3. recalc. distance matrix w. new group
4. repeat until tree is finished - 131What’s the assumption made by UPGMA & problem?Assume: constant rate of mutation
Problem: rate heterogenity betwn gene & species; also diff generation time - 132What’s the purpose of neighbor-joining method & how does it work?Goal: minimize total branch length of tree
Take pairwise distance of taxons & compare distance to rest of tree, merge, recalc. distance matrix until tree is made - 133Roughly outline the math for NJ trees.1. for each node, calc. approx. dist. to rest of tree
2. join i,j so branch length minimized
3. calc node i,j to (i,j)
4. calc. distance of other node D(i,j)k
5. add (i,j) to table & regenerate DM until are nodes are combined - 134How is the branch length at each step minimized? (IMPORTANT)By considering distance of node to rest of the tree
- 135Name 6 properties related to NJ method.1. greedy algorithm
2. can reproduce tree topology if branch lengths given
3. fits well w. stats model of evolution
4. acc. for branch length variation (evol. rate)
5. can produce - branch length
6. can lose info in distance matrix - 136What does bootstrap measure and its purpose? How does it work?Measures confidence for phylogenetic hypothesis.
Make MSA > divide to N-ordered sites > sample N sites from alignmt w. replacemt > recalc. tree & det. freq. of each node (clade) in each rep - 137What are some advantages and disadvantages of bootstrapping?Useful for interpreting robustness of tree based on data given & give better representation than just best tree
- 138What is a pseudoreplication?Sampling own data with replacement to see if data is reproduced as a measure of confidence
- 139What is a character? What does this mean in the context of molecular phylogenies?Any attribute of organism that can provide insight to shared ancestry
Molecular phylogeny = NT position in a gene - 140What’s the difference between a plesiomorphy vs. apomorphy?Plesiomorphy = ancestral character state
Apomorphy = derived character state (DIFF from ancestor) - 141What is a synapomorphy and autapomorphy? What are they under?Synapo.: derived character state SHARED by >=2 taxa via inheritance from common ancestor (informative)
Autapomorphy: uniquely derived character state (uninformative) - 142What is monophyletic?Group that includes ALL descendants of common ancestor (clades)
- 143What are the two cases that demonstrate non-monophyletic groups?Paraphyletic - gp that includes some but not al members of descendants
Polyphyletic - tax errorneously grouped together based on homoplasy - 144Specific to polyphyletic taxa, what is a key trait?Homoplasy - shared character but NOT FROM COMMON ANCESTOR (ex. parallel evolution/convergent evolution)
- 145What does the parsimony method do wrt to phylogeny? What are the assumptions?Look for tree that require fewest # of changes.
Assume mutations are rare events, so fewer means model more likely correct - 146What are informative vs. uninformative sites wrt parsimony?Informative - site that favor a specific tree; has 2 nt that appears >= 2 times
Uninformative - doesn't favor a tree (ex. invariant sites) - 147What’s a major problem associated with considering all trees & keeping those with fewest mutations?Very computationally intensive and not feasible
- 148What’s a generalized approach to explore tree space?1. Make quick n dirty starting tree (ex. NJ)
2. make change to tree via moving branch/nodes
3. keep trees w. improved score over time - 149Name the 5 approaches used to explore tree space?1. Branch and Bound
2. Nearest neighbor interchange
3. Subtree pruning & regrafting
4. Tree bisection & reconnection
5. Parsimony ratchet - 150What’s the strategy for faster searches and how does it work?Branch & Bound
Find upper L for tree > build tree by adding one branch at a time > if exceed bound L, stop tree & eliminate > find lower bound L' for completed & use it as new value > continue until only 1 tree remain - 151Briefly describe how a nearest neighbor interchange works (NNI)Swaps internal branch of tree & keep if new score improves
- 152Briefly describe how a nearest neighbor interchange works (NNI)Subtree cut from original tree & moved to new location; keep if score improves
- 153Describe how the tree bisection and reconnection (TBR) worksSplit starting tree to 2
Edge combination for subtree recombined to generate new tree set
Keep if score improves - 154What is one major issue with tree exploration? Solution?Getting stuck in local minima
Parsimony ratchet - get a starting tree, reweigh a subset of MSA alignmt, explore new tree space & keep best, reset original wt & explore tree space again > keep best tree & repeat many times - 155Name 4 reasons why phylogeny is important?1. help understand evolution
2. implications for medicine
3. implication for agriculture
4. implication for ecology/env - 156What are the 3 criteria that should apply to tree reconstructions?1. need accurate MSA
2. compare diff model of evolution
3. assess tree quality - 157What are possible factors to consider in models of evolution?1. substitution preference (ex. transition vs. transversion)
2. rate of substitution at diff seq position
3. diff rates on diff tree branch (i.e. rate heterogeneity) - 158Name the 2 measures of evolutionary distance.1. fractional alignmt distance (p-distance)
2. poisson distance - 159What’s the p-distance (formula); associated problems (3) ?p= D/L
Estimate probability of site being mutated
Problem: 1) statistical variation in short seq, 2) doesn't acc for multiple substitution, 3) doesn't acc. for diff rate betwn seq. - 160How is Poisson distance different from p-distance? Assumption?Acc for multiple sub at individual site & assume fixed rate of mutation per time
- 161What’s the goal of nucleotide models?Represent nt changes in set of seqs
- 162What are the difference between Jukes & Cantor, Kimura and Tamura models?JC = equal nt composition, single parameter model, reversible
Kimura = 2 parameter model, equal nt composition
Tamura = multi-parameter model, variable nt composition - 163What are the 3 assumptions made by Jukes-Cantor model?1. all sites independent
2. rates of evolution same at all sites
3. all substitution equally likely, at rate alpha - 164What does K2P model account for?Acc for substitution preferences (i.e. transition vs. transversion); transversion expected to occur at lower rate
- 165What’s the commonality between JC and K2P models?Both assume 1:1:1:1 nt composition
- 166What are the differences between JC, K2P and REV models?JC = 1 parameter
K2P = 2 aprameter
GTR = multiple parameter, irreversible - 167How can different rates at different positions be corrected? Give ex. of this case?Gamma distribution
ex. JC model modification - 168What is the common model used for protein evolution? Name 3 common models?Use empirically derived substitution matrices
ex. JTT, WAG, LG - 169What’s the effect of gamma parameter a on distribution?Low a = some nt undergo slow sub; most variation at few nt sites that evolve rapidly (HIGH VARIABILITY); asymptotic
Large a = few nt site undergo very fast/slow evolution (LOW VARIABILITY); peak - 170What is used to compare the models & det the quality? How is the model selected?Use goodness-of-fit & model complexity
Akaike Information Criterion
Use AIC = -2ln L + 2k
L= log likelihood (fit)
k = # of free parameters (complexity)
Want to minimize AIC so get model with highest likelihood & least # of parameters