Comparing strings online

Regions of similarity in two genetic strings that are presumed to be homologous can be found by alignment.

An alignment of two human zinc finger proteins, identified by their GenBank accession numbers on the left. An asterisk (*) below aligned symbols denotes identical amino acids, a colon (:) denotes similar amino acids, a dot (.) denotes less similar amino acids, and a blank space denotes dissimilar amino acids. Color-coding represents the amino acid's chemical properties (e.g., blue denotes acidic and green denotes basic).

As shown in the above figure, a global alignment of two strings is formed by adding gap symbols (-) to each string to make them the same length, so that every symbol in one string corresponds to a symbol in the other. A gap symbol in an aligned string represents a deletion at that position or an insertion in the string it is aligned to. Mismatches can be interpreted as point mutations that change symbols at single positions in the strings.

There are many possible alignments for a pair of strings. In practice, we cannot know which is the absolute "best" alignment of two genetic strings. This is because discriminating between alignments requires assumptions about how the strings have evolved, and this information about their evolution is precisely what we hope to learn from the alignment. The best we can do is to compare alignments quantitatively by using a scoring approach. Many such alignment scores assign 0 to a match, a negative number to a mismatch, and an even larger negative number to a gap. Summing all these scores over an alignment gives the alignment's score. In keeping with the assumption of parsimony, the optimal alignment will have the highest score, implying the fewest possible changes between the strings.

A scoring matrix contains a score for matches and mismatches between each pair of symbols. These scores are derived empirically based on the frequency with which these matches and mismatches occur in highly similar biological sequences for which the true alignment is practically unequivocal.

A gap penalty is the score deducted for the presence of a gap. One of the most commonly used gap penalties is an affine gap penalty, in which one penalty is assigned for adding the first symbol to a gap and different penalty is charged for every additional symbol in the gap. The gap opening penalty is usually larger than the gap extension penalty because a single deletion of $$k$$ nucleotides is more likely than $$k$$ independent, contiguous, single-nucleotide deletions.

For example, say that we use a simple scoring matrix that rewards $$+1$$ for all matching symbols and charges $$-1$$ for all mismatches, along with an affine gap penalty that has a gap opening penalty of $$-5$$ and a gap extension penalty of $$-1$$. The following alignment has 14 matched symbols, 3 mismatches, and two gaps (one of which is extended by three symbols), which gives a total score for the alignment of $$14 - 3 - (2\cdot 5 + 3) = -2$$.

GARFIELDTHEVERYFA-TCAT
:::||||||||    || ||||
WINFIELDTHE----FASTCAT

If you would like to learn about the algorithmic ideas behind the global alignment algorithms, you may be interested in solving "Global alignment with scoring matrix" and "Global alignment with scoring matrix and affine gap penalty".

EBI hosts many different online user interfaces for bioinformatics tools, including Needle and Stretcher. Needle and Stretcher are tools from the EMBOSS package that are for aligning genetic strings. Both can be accessed from here¹.

The programs are very similar. One difference is that Stretcher always uses an end gap penalty, while Needle includes the option but does not use it by default. In either of these interfaces, you can enter strings directly or in any supported format: GCG, FASTA, EMBL, GenBank, PIR/NBRF, Phylip, or the UniProtKB/Swiss-Prot format. Feel free to play around with these interfaces by entering pairs of strings. Notice that clicking on More options… under "Step 2" will allow you to change the alignment parameters as well as the output format.

Assignment

An online interface to EMBOSS's Needle tool for aligning DNA and RNA strings can be found here².

Use:

the DNAfull scoring matrix; note that DNAfull uses IUPAC notation for ambiguous nucleotides
gap opening penalty of 10
gap extension penalty of 1

For our purposes, the "pair" output format will work fine. This format shows the two strings aligned at the bottom of the output file beneath some statistics about the alignment.

Write a function globalAlignmentScore that takes two GenBank IDs. The function must return the maximum global alignment score between the DNA strings associated with these IDs.

Example

        >>> globalAlignmentScore('JX205496.1', 'JX469991.1')
257

Programming shortcut

You can download the EMBOSS suite from the EMBOSS homepage on Sourceforge and run Needle and Stretcher locally via the command line (or by using a graphical user interface, such as Jemboss for UNIX systems).

The Needle help page contains detailed descriptions of command line arguments. Do not forget to count gaps on the ends of an alignment using the arguments -endweight, -endopen and -endextend.