Transcript Bioinformatics Lecture 1

Bioinformatics:
Copyright© Kerstin Wagner
Introduction: What is bioinformatics?
Can be defined as the body of tools, algorithms needed to handle large
and complex biological information.
Bioinformatics is a scientific discipline created from the interaction
of biology and computer science.
The NCBI defines bioinformatics as:
"Bioinformatics is the field of science in which biology, computer
science, and information technology merge into a single discipline”
Genomics era: High-throughput DNA sequencing
The first high-throughput genomics
technology was automated DNA sequencing
in the early 1990.
In 1995, Venter and Hamilton used whole-
genome shotgun sequencing strategy to
sequence the genomes of Mycoplasma and
Haemophilus .
In September 1999, Celera Genomics
completed the sequencing of the
Drosophila genome.
The 3-billion-bp human genome sequence
was generated in a competition between
the publicly funded Human Genome
Project and Celera
High-throughput DNA sequencing
Top image: confocal detection
by the MegaBACE sequencer
of fluorescently labeled DNA
That was then. How about
now?
Next Generation Sequencing
(2010) vol11:31
Genomics: Completed genomes as of 2013
Currently the genome of the organisms are sequenced:
6131 bacterial/146 archaeal/166 eukaryotic (3900 organelles)
This generates large amounts of information to be handled by individual
computers.
The trend of data growth

century is a century of biotechnology and OMICS:
Genomics: New sequence information is being
produced at increasing rates. (The
contents of GenBank double every year)
Nucleotides(billion)
21st
8
7
6
5
4
3
2
1
0
1980

1985
Transcriptomics: Microarray: Global expression analysis: RNA
levels of every gene in the genome analyzed in parallel.
Progressively replaced by RNA-seq

Proteomics: Global protein analysis generates by large mass
spectra libraries.

Metabolomics: Global metabolite analysis: 25,000 secondary
metabolites characterized
1990
Years
1995
2000
How to handle the large amount of information?
Drew Sheneman, New Jersey--The Newark Star Ledger
Answer: bioinformatics and Internet
Bioinformatics history
In1960s:
the birth of bioinformatics
IBM 7090 computer
Margaret
Oakley Dayhoff created:
The first protein database
The first program for sequence assembly

There
is a need for computers and algorithms that allow:
Access, processing, storing, sharing, retrieving, visualizing, annotating…
Why do we need the Internet?
“omics”
projects and the information associated with involve a huge amount
of data that is stored on computers all over the world.
Because
it is impossible to maintain up-to-date copies of all relevant
databases within the lab. Access to the data is via the internet.
Database
storage
You are
here
The Commercial Market
Current
bioinformatics market is worth 300 million / year
(Half software)
Prediction: $2
~50
billion / year in 5-6 years
Bioinformatics companies:
Genomatrix Software, Genaissance Pharmaceuticals, Lynx, Lexicon Genetics, DeCode
Genetics, CuraGen, AlphaGene, Bionavigation, Pangene, InforMax, TimeLogic,
GeneCodes, LabOnWeb.com, Darwin, Celera, Incyte, BioResearch Online, BioTools,
Oxford Molecular, Genomica, NetGenics, Rosetta, Lion BioScience, DoubleTwist,
eBioinformatics, Prospect Genomics, Neomorphic, Molecular Mining, GeneLogic,
GeneFormatics, Molecular Simulations, Bioinformatics Solutions….
Scope of this lab
The
lab will touch on the following computational tasks:
Similarity
search
Sequence comparison: Alignment, multiple alignment, retrieval
Sequences analysis: Signal peptide, transmembrane domain,…
Protein folding: secondary structure from sequence
Sequence evolution: phylogenetic trees
Make
you familiar with bioinformatics resources available on the
web to do these tasks.
Applying algorithms to analyze genomics data
-Accession #?
-Annotation?
Is it already in
databases?
Protein
characteristics?
-Sub-localization
-Soluble?
-3D fold
Is there conserved
regions?
-Alignments?
-Domains?
Other
information?
You have just
cloned a gene
Is there similar
sequences?
-% identity?
-Family member?
-Expression profile?
-Mutants?
Evolutionary
relationship?
-Phylogenetic
tree
A critical failure of current bioinformatics is the lack of a single software
package that can perform all of these functions.
DNA (nucleotide sequences) databases
They
are big databases and searching either one should produce
similar results because they exchange information routinely.
-GenBank (NCBI): http://www.ncbi.nlm.nih.gov
-Ensembl: http://useast.ensembl.org/index.html
-DDBJ (DNA DataBase of Japan): http://www.ddbj.nig.ac.jp
-TIGR: http://tigr.org/tdb/tgi
-Yeast: http://yeastgenome.org
-Microbes: http://img.jgi.doe.gov/cgi-bin/pub/main.cgi
Protein (amino acid) databases
Known
proteins:
-Swiss-Prot (very high level of annotation)
http://au.expasy.org/
-PIR (protein identification resource) the world's most
comprehensive catalog of information on proteins
http://www.pir.uniprot.org/
Translated
databases:
-TREMBL (translated EMBL): includes entries that have
not been annotated yet into Swiss-Prot.
http://www.ebi.ac.uk/trembl/access.html
-GenPept (translation of coding regions in GenBank)
-pdb (sequences derived from the 3D structure
Brookhaven PDB) http://www.rcsb.org/pdb/
Database homology searching
Use
algorithms to efficiently provide mathematical basis of searches
that can be translated to statistical significance.
Assumes
that sequence, structure, and function are inter-related.
All
similarity searching methods rely on the concepts of alignment
and distance between sequences.
A similarity
score is calculated from a distance: the number of DNA
bases or amino acids that are different between two sequences.
Calculating alignment scores
Scoring
system: Uses scoring matrices that allow biologists to quantify the
quality of sequence alignments.
The
raw score S is calculated by summing the scores for each aligned
position and the scores for gaps. Gap creation/extension scores are
inherent to the scoring system in use (BLAST, FASTA…)
The
score for an identity or a mismatch is given by the specified substitution
matrix (e.g., BLOSUM62).
Devising a scoring system
Some
popular scoring matrices are:


How
PAM (Percent Accepted Mutation): for evolutionary studies.
For example in PAM1, 1 accepted point mutation per 100 amino
acids is required.
BLOSUM (BLOcks amino acid SUbstitution Matrix): for finding
common motifs. For example in BLOSUM62, the alignment is
created using sequences sharing no more than 62% identity.
the matrices were created:

Very similar sequences were aligned.
From these alignments, the frequency of substitution between
each pair of amino acids was calculated and then PAM1 was built.

After normalizing to log-odds format, the full series of PAM matrices
can be calculated by multiplying the PAM1 matrix by itself.

Devising a scoring system
Importance:
Scoring matrices appear in all analysis
involving sequence comparison.

The choice of matrix can strongly influence
the outcome of the analysis.

Understanding theories underlying a given
scoring matrix can aid in making proper
choice:
-Some matrices reflect similarity: good for
database searching

-Some reflect distance: good for phylogenies
Database search methods: Sequence Alignment
Two
broad classes of sequence alignments exist:
QKESGPSSSYC

Global alignment:
not sensitive
VQQESGLVRTTC
ESG

Local alignment:
faster
ESG
The
most widely used local similarity algorithms are:
Smith-Waterman (http://www.ebi.ac.uk/MPsrch/)
Basic Local Alignment Search Tool (BLAST, http://www.ncbi.nih.gov)
Fast Alignment (FASTA, http://fasta.genome.jp; http://www.ebi.ac.uk/fasta33/;

http://www.arabidopsis.org/cgi-bin/fasta/nph-TAIRfasta.pl)
Which algorithm to use for database similarity search?
Speed:
BLAST > FASTA > Smith-Waterman (It is VERY SLOW and uses a
LOT OF COMPUTER POWER)

Sensitivity/statistics:
FASTA is more sensitive, misses less homologues
Smith-Waterman is even more sensitive.
BLAST calculates probabilities
FASTA more accurate for DNA-DNA search then BLAST

-tuple methods provide optimal alignments
These
methods are faster and excellent in comparing sequences.
BLAST
and FASTA programs are based on -tuple algorithms:
1-Using query sequence, derive a list of
words of length w (e.g., 3)
2-Keep high-scoring words using a
scoring matrix(e.g. BLOSUM 62)
3-High-scoring words are compared
with database sequences
4-Sequences with many matches to
high-scoring words are used for final
alignments
Tools to search databases
The
dilemma: DNA or protein?
Search by similarity
Using nucleotide seq.



Using amino acid seq.
Is the comparison of two nucleotide sequences accurate?
By translating into amino acid sequence, are we losing information?
The genetic code is degenerate (Two or more codons can represent
the same amino acid)
Very different DNA sequences may code for similar protein sequences
We certainly do not want to miss those cases!
Reasons for translating
Comparing
DNA sequences give more random matches:
A good alignment with end-gaps
A very poor alignment
Almost 50% identity!
Conservation
of protein in evolution (DNA similarity decays faster!)
Conclusion:
It is almost always better to compare coding sequences in their amino acid form,
especially if they are very divergent.


Very highly similar nucleotide sequences may give better results.
BLAST and FASTA variants
FASTA:
Compares a DNA query to DNA database, or a protein query
to protein database
FASTX:
Compares a translated DNA query to a protein database
TFASTA: Compares a protein query to a translated DNA database
BLASTN:
Compares a DNA query to DNA database.
BLASTP:
Compares a protein query to protein database.
BLASTX:
Compares the 6-frame translations of DNA query to protein
database.
Compares a protein query to the 6-frame translations of a DNA
database. You can however define your frame of interest
TBLASTN:
TBLASTX:
Compares the 6-frame translations of DNA query to the 6-frame
translations of a DNA database (each sequence is comparable to
BLASTP searches!)
PSI-BLAST: Performs iterative database searches. The results from each round
are incorporated into a 'position specific' score matrix, which is
used for further searching
A practical example of sequence alignment
http://www.ncbi.nlm.nih.gov
BLAST results
Detailed BLAST results
E
value: is the expectation value or probability to find by chance hits similar to
your sequence. The lower the E, the more significant the score.
Database searching tips
Use
latest database version.
Use
BLAST first, then a finer tool (FASTA,…)
Search
both strands when using FASTA.
Translate
Search
sequences where relevant
6-frame translation of DNA database
E
< 0.05 is statistically significant, usually biologically
interesting.
If
the query has repeated segments, delete them and
repeat search
Most widely used sites for sequence analysis
Sites
for DNA to protein translation:
These algorithms can translate DNA sequences in any of the 3 forward or three
reverse sense frames.
Translate (http://au.expasy.org/tools/dna.html)
Translate a DNA sequence: (http://www.vivo.colostate.edu/molkit/translate/index.html)
Transeq (http://www.ebi.ac.uk/emboss/transeq)
Sites
for alignment of 2 sequences:
T-COFFEE (http://tcoffee.vital-it.ch/cgi-bin/Tcoffee/tcoffee_cgi/index.cgi): more
accurate than ClustalW for sequences with less than 30% identity.
ClustalW (http://www.ch.embnet.org/software/ClustalW.html;

http://align.genome.jp)
bl2sequ (http://www.ncbi.nlm.nih.gov/blast/bl2seq/wblast2.cgi)
LALIGN (http://www.ch.embnet.org/software/LALIGN_form.html)
MultiALIGN (http://prodes.toulouse.inra.fr/multalin/multalin.html)

BioEdit — a sequence editing software package
http://www.mbio.ncsu.edu/bioedit/bioedit.html
Oligo Design and Analysis Tools
IDT: http://www.idtdna.com/scitools/scitools.aspx
Primer3: http://frodo.wi.mit.edu/primer3/ or directly on NCBI