Bioinformatics Tool: Multiple Sequence Alignment for Analyzing Biological Sequences, Study notes of Chemistry

Download Bioinformatics Tool: Multiple Sequence Alignment for Analyzing Biological Sequences and more Study notes Chemistry in PDF only on Docsity!

Multiple Sequence Alignment

BME 110: CompBio Tools Todd Lowe April 22, 2008

Multiple Sequence Alignment

Multiple sequence alignment is probably the single-most important bioinformatics tools.

Many applications require accurate MSAs - PSIBLAST - Family and domain classification - Pattern identification - Structure prediction - secondary structure • fold recognition - Phylogeny - Full-genome alignments in browsers

PSI-BLAST Alignments

The goal of BLAST is rapid detection bydetecting high-scoring local alignments. Itdoesn’t necessarily find the optimal global orlocal alignment

Profiles throw away information for regions thatare insertions relative to the query

Methods

Full Dynamic Programming - Gives the optimal solution, but prohibitively slow and memory intensive for >6-8 sequences - MSA program - Progressive Alignment - ClustalW - http://www.ebi.ac.uk/clustalw/index.html (most commonly used) - Tcoffee - http://igs-server.cnrs-mrs.fr/Tcoffee/ (a little better, but slower) - Iterative - better than progressive methods, but slower - Dialign - HMMs

ClustalW Example

Input: 5 sequences detected by BLASTp using human SNAP-25as a query - Default parameters, output order: input >sp_P13795 MAEDADMRNELEEMQRRADQLADESLESTRRMLQLVEESKDAGIRTLVMLDEQGEQLERIEEGMDQINKD MKEAEKNLTDLGKFCGLCVCPCNKLKSSDAYKKAWGNNQDGVVASQPARVVDEREQMAISGGFIRRVTND ARENEMDENLEQVSGIIGNLRHMALDMGNEIDTQNRQIDRIMEKADSNKTRIDEANQRATKMLGSG >gi_31242623 MPAAAPPAENGAAVPKTELQELQMKQQQVVDESLDSTRRMLALCEESTEVGMRTIVMLDEQGEQLDRIEE GMDQINADMREAEKNLSGMEKCCGICVLPCNKSASFKEDDGTWKGNDDGKVVNNQPQRVMDDRNGLGPQA GYIGRITNDAREDEMEENMGQVNTMIGNLRNMALDMGSELENQNRQIDRINRKGDSNATRIAAANERAHD LLK >gi_3822409 MPTTAEPAQENGAPRSELQELQLKAGQVTDETLESTRRMLALCEESKEAGIRTLVALDDQGEQLERIEEN MDQINADMKEAEKNLTGMEKFCGLCVLPWNKSAPFKENEDAWKGNDDGKVVNNQPQRVMDDGSGLGPQGG YIGRITNDAREDEMEENVGQVNTMIGNLRNMAIDMGSELENQNRQIDRIKNKAEM >gi_39593308 MSARRGAPGGQRHPRPYAVEPTVDINGLVLPADMSDELKGLNVGIDEKTIESLESTRRMLALCEESKEAG IKTLVMLDDQGEQLERCEGALDTINQDMKEAEDHLKGMEKCCGLCVLPWNKTDDFEKNSEYAKAWKKDDD GGVISDQPRITVGDPTMGPQGGYITKITNDAREDEMDENIQQVSTMVGNLRNMAIDMSTEVSNQNRQLDR IHDKAQSNEVRVESANKRAKNLITK >gi_32567202 MSGDDDIPEGLEAINLKMNATTDDSLESTRRMLALCEESKEAGIKTLVMLDDQGEQLERCEGALDTINQD MKEAEDHLKGMEKCCGLCVLPWNKTDDFEKTEFAKAWKKDDDGGVISDQPRITVGDSSMGPQGGYITKIT NDAREDEMDENVQQVSTMVGNLRNMAIDMSTEVSNQNRQLDRIHDKAQSNEVRVESANKRAKNLITK

Input Formats

FASTA format

Download from NCBI, ExPASy, EBI, Pfam … - Sequence names should be - Unique - 15 characters or less - Comprised of only A-Z,a-z,0-9 and _ (Do not use #$%@|*!:;. or spaces)

ClustalW Guide Tree

The guide tree shows the distances between sequences obtained fromthe initial pairwise alignments - This is the order that sequences were added into the MSA - Guide tree is not a phylogenetic tree (it’s just a rough estimate of similarity), however a true phylogenetic tree can be generated aftermaking an alignment

Progressive Alignment

Greedy algorithm

Breaks problem up into smaller problems - Finds best solution to each small problem - Combine solutions to get answer to whole problem - Not necessarily the global answer - Doesn’t use all information in solving sub-problems - Suboptimal answers for small problems may combine togive a better overall answer - Gaps: once created, they stay as part of alignmentfor rest of alignment iterations

Interleaved Formats

Most common output formats for MSAs areinterleaved:

MSF, ASN, BLAST query-anchored formats - All sequences are stacked up, and chopped intoblocks of ~60 residues - Easy for humans to read, but difficult to edit - Tools for converting formats are available on theweb ReadSeq at Baylor College of Medicine HGSC: http://searchlauncher.bcm.tmc.edu/seq-util/Options/readseq.html

Aligned FASTA Format

SN29_RAT/142-196 PSSRLKEAINTSKDQESKYQASHPNLRRLHDAE---LDSVPASTV----NTEVY-----P KNSSL---R-----A >SN29_HUMAN/142-197 PNNRLKEAISTSKEQEAKYQASHPNLR-------KLDDTDPVPRGA---GSAMSTDA-YP KNPHL---R-----A >SN25_TORMA/95-148 PCNK----LKNFEAGGAYKKVWGNNQD------G-VVASQP-ARVMD-DREQMA-----M SGGYI--RRI-TDDA >O93578/11-59 PCNK----MKS-----GASKAWGNNQD------G-VVASQP-ARVVD-EREQMA-----I SGGFI--RRV-TDDA >SN25_DROME/98-149 PCNK----SQSFK---EDDGTWKGNDD------GKVVNNQP-QRVMD-DRNGM-----MA QAGYI--GRI-TNDA • Uppercase and ‘-’ characters are alignment columns. There must be thesame number of aligned characters in all sequences.

Insertions that are not part of the alignment, are indicated with lowercase and ‘.’ characters. These are not read (i.e. they’re for humansonly) - Benefits - Easily machine readable - Readable by most programs that read FASTA format

Sequence Logos

Logos are another useful visualization of alignments that allowconserved positions to be easily picked out. - Multiple tools available on the web or can be downloaded: - http://weblogo.berkeley.edu

Tcoffee

Makes a library of pair-wise global and severallocal alignments

Tries to find a multiple alignment that has bestconsensus with all alignments in the library. - Still a progressive algorithm - Slower, but usually a bit better than ClustalW

Criteria for a Good MSA

Most methods align proteins on the basis of sequencesimilarity, but what we really want to know is: - Evolutionary similarity - Functional similarity - Structural similarity - If the sequences are closely related, these similarities areall equivalent. As sequences become more divergent,theses similarities may not be equivalent. - There isn’t necessarily one ‘correct’ alignment for a family.MSA doesn’t necessarily reflect a true structural or functional alignment.

Which Sequences?

Don’t include too many

Problems are VERY slow for many sequences - Start with 10-15 or so. - Closely related sequences are easy to align, butless informative. The converse is true for moredistantly related sequences - No identical sequences - Each sequence 30-70% identical with at least half of theother sequences.