Transcript Sequencing genomes

Last lecture summary
• identity vs. similarity
• homology vs. similarity
• gap penalty
• affine gap penalty
• gap penalty high
• fewer gaps, if investigating related sequences
• low
• more gaps, larger gaps, distantly related sequences
BLOSUM
• blocks
• focuse on substitution patterns only in blocks
• BLOSUM62 – 62, what does it mean?
• BLOSUM vs. PAM
• BLOSUM matrices are based on observed alignments
• BLOSUM numbering system goes in reversing order as the PAM
numbering system
Selecting an Appropriate Matrix
Matrix
Best use
Similarity (%)
Pam40
Short highly similar alignments
70-90
PAM160
Detecting members of a protein family
50-60
PAM250
Longer alingments of more divergent sequences
~30
BLOSUM90
Short highly similar alignments
70-90
BLOSUM80
Detecting members of a protein family
50-60
BLOSUM62
Most effective in finding all potential similarities
30-40
BLOSUM30
Longer alingments of more divergent sequences
<30
Similarity column gives range of similarities that the matrix is able to best detect.
Dynamic programming (DP)
• Recursive approach, sequential dependency.
• 4th piece can be solved using solution of the 3rd
piece, the 3rd piece can be solved by using solution of
the 2nd piece and so on…
New best alignment = previous best + local best
Best previous alignment
Sequence A
...
...
...
...
Sequence B
If you already have the optimal solution to:
X…Y
A…B
then you know the next pair of characters will either be:
X…YZ
A…BC
or
X…YA…BC
or
X…YZ
A…B-
You can extend the match by determining which of these
has the highest score.
• Window size?
• Stringency?
• Color mapping?
Frame shifts?
New stuff
Homology vs. similarity again
• Just a reminder of the important concept in sequence
analysis – homology. It is a conclusion about a common
ancestral relationship drawn from sequence similarity.
• Sequence similarity is a direct result of observation from
the sequence alignment. It can be quantified using
percentages, but homology can not!
• It is important to understand this difference between
homology and similarity.
• If the similarity is high enough, a common evolutionary
relationship can be inferred.
Limits of the alignment detection
• However, what is enough? What are the detection limits of
pairwise alignments? How many mutations can occur
before the differences make two sequences
unrecognizable?
• Intuitively, at some point are two homologous sequences
too divergent for their alignment to be recognized as
significant.
• The best way to determine detection limits of pairwise
alignment is to use statistical hypothesis testing. See
later.
Twilight zone
• However, the level one can infer homologous relationship
depends on type of sequence (proteins, NA) and on the
length of the alignment.
• Unrelated sequences of DNA have at least 25% chance to be
identical. For proteins it is 5%. If gaps are allowed, this percentage
can increase up to 10-20%.
• The shorter the sequence, the higher the chance that some
alignment can be attributed to random chance.
• This suggest that shorter sequences require higher cuttof
for inferring homology than longer sequences.
Essential bioinformatics, Xiong
Statistical significance
• Key question – Constitutes a given alignment evidence for
homology? Or did it occur just by chance?
• The statistical significance of the alignment (i.e. its score)
can be tested by statistical hypotheses testing.
Significance of global alignment
• We align two proteins: human beta globin and myoglobin.
We obtain score S.
• And we want to know if such a score is significant or if it
appeared just by a chance. How to proceed?
• State H0
• two sequences are not related, score S represents a chance
occurrence
• State Ha
• Choose a significance level 𝛼
• Statistics of distribution.
• i.e. sample mean, sample standard deviation
Database similarity searching
BLAST
• Basic Local Alignment Search Tool (BLAST) – Google of the
sequence world.
• Compare a protein or DNA sequence to other sequences in
various databases, main tool of NCBI.
• Why to search database
• Determine what orthologs and paralogs are known for a particular
•
•
•
•
•
sequence.
Determine what proteins or genes are present in a particular organism.
Determine the identity of a DNA or protein sequence.
Determine what variants have been described for a particular gene or
protein.
Investigate ESTs.
Explore amino acid residues that are important in the function and/or
structure of a protein (multiple alignment of BLAST results, conserved
residues).
Database searching requirements I
• query sequence, perform pairwise alignments between
•
•
•
•
the query and the whole database (target)
Typically, this means that millions of alignments are
analyzed in a BLAST search, and only the most closely
related matches are returned.
We are usually more interested in identifying locally
matching regions such as protein domains. Global
alignment (Needlman-Wunsch) is not often used.
Smith-Watermann is too computationally intensive.
Instead, heuristic is utilized, significant speed up.
Database searching requirements II
• sensitivity – the ability to find as many correct hits (TP)
as possible
• selectivity (specificity) – ability to exclude incorrect hits
(FP)
• speed
• ideally: high sensitivity, high specificity, high speed
• reality: increase in sensitivity leads to decrease in
specificity, improvement in speed often comes at the cost
of lowered sensitivity and selectivity
Types of algorithms
• exhaustive
• uses a rigorous algorithm to find the exact solution for a particular
problem by examining all mathematical combinations
• example: DP
• heuristic
• computational strategy to find an empirical or near optimal solution
by using rules of thumb
• this type of algorithms take shortcuts by reducing the search space
according to some criteria
• the shortcut strategy is not guaranteed to find the best or most
accurate solution
Heuristic algorithms
• Perform faster searches because they examine only a
fraction of the possible alignments examined in regular
dynamic programming
• currently, there are two major algorithms:
• FASTA
• BLAST
• Not guaranteed to find the optimal alignment or true
homologs, but are 50–100 times faster than DP.
• The increased computational speed comes at a moderate
expense of sensitivity and specificity of the search, which
is easily tolerated by working molecular biologists.
BLAST
• Parts of algorithm
• list, scan, extend
• BLAST uses word method for pairwise alignment
• Find short stretches of identical (or nearly identical) letters
in two sequences – words (similar to window in dot plot)
• Basic assumption: two related sequences must have at
least one word in common
• By first identifying word matches, a longer alignment can
be obtained by extending similarity regions from the
words.
• Once regions of high sequence similarity are found,
adjacent high-scoring regions can be joined into a full
alignment.
BLAST - list
• Compile a list of “words” of a fixed length w that are
•
•
•
•
derived from the query sequence.
protein searches – word size = 3, NA searches = 11
A threshold value T is established for the score of aligned
words (true for proteins, for NAs exact matches are used).
Those words either at or above the threshold are
collected and used to identify database matches; those
words below threshold are not further pursued.
The threshold score T can be lowered to identify more
initial pairwise alignments. This will increase the time
required to perform the search and may increase the
sensitivity
BLAST - scan
• After compiling a list of word pairs at or above threshold T,
the BLAST algorithm scans a database for hits.
• This requires BLAST to search an index of the database
to find entries that correspond to words on the compiled
list.
BLAST - extend
• Extend hits to find alignments called high-scoring
segment pairs (HSPs).
• Extend in both directions (ungapped originally, gapped
BLAST is newer), count the alignment score.
• The extension process is terminated when a score falls
below a cutoff.
BLAST strategy
• Compare a protein or DNA query sequence to each
database entry and form pairwise alignments (HSPs).
• When the threshold parameter is raised, the speed of the
search is increased, but fewer hits are registered, and so
distantly related database matches may be missed.
• When the threshold parameter is lowered, the search
proceeds more slowly, but many more word hits are
evaluated, and thus sensitivity is increased.
• Recent improvement – gapped BLAST
• Variants
• BLASTN – nucleotide sequences
• BLASTP – protein sequences
• BLASTX – uses nucleotide sequences as queries and translates
them in all six reading frames to produce translated protein
sequences, which are used to query a protein sequence database
• TBLASTN – queries protein sequences to a nucleotide sequence
database with the sequences translated in all six reading frames
• TBLASTX – uses nucleotide sequences, which are translated in all
six frames, to search against a nucleotide sequence database that
has all the sequences translated in six frames.
Which sequence to search?
• The choice of the type of sequences also influences the
sensitivity of the search.
• Clear advantage of using protein sequences in detecting
homologs
• If the input sequence is a protein-encoding DNA sequence, use
BLASTX (six open reading frames before sequence comparisons)
• If you’re looking for protein homologs encoded in newly
sequenced genomes, you may use TBLASTN. This may
help to identify protein coding genes that have not yet
been annotated.
• If a DNA sequence is to be used as the query, a proteinlevel comparison can be done with TBLASTX.
• TBLASTN, TBLASTX are very computationally intensive
and the search process can be very slow.
E-value I
• expected value
• a parameter that describes the number of hits one can
'expect' to see by chance when searching a database of a
particul
• decreases exponentially as the score of the match
increasesar size
• an E value of 1 assigned to a hit can be interpreted as
meaning that in a database of the current size one might
expect to see 1 match with a similar score simply by
chance
• If the database were twice as big, there would be twice the
likelihood of finding a score equal to or greater than S by chance.
E-value II
• E < 10-50 … extremely high confidence that the database match
is a result of homologous relationships
• E is from (10-50 , 0.01) … the match can be considered a result
of homology (for proteins, conclusive are E-values < 0.001)
• E is from (0.01, 10) … the match is considered not significant,
but may hint tentative remote homology
• E > 10 … the sequences under consideration are either
unrelated or related by extremely distant relationships that fall
below the limit of detection with the current method.
• E-value is proportional to the database size, as database
grows E-value for a given sequence match increases.
However, the evolutionary relationship between two sequences
remains constant. As the db grows, one may lose previously
detected homologs.
Bit score
• A typical BLAST output reports both E values and scores.
• There are two kinds of scores: raw and bit scores.
• Raw scores are calculated from the substitution matrix and the gap
penalty parameters that are chosen.
• The bit score S’ is calculated from the raw score by normalizing
with the statistical variables that define a given scoring system.
• Bit scores from different alignments, even those
employing different scoring matrices in separate BLAST
searches, can be compared.