For more info on the purpose of this code, please check out this blog post.

So, how do you use this anyways?

Test it out:

First compile it - I'm not sure how portable my makefile is, but as long as you have g++ and python you shouldn't have any problems. If you do run into problems, or have any portability tricks to share, let me know.

make

Then get some test data. This script uses the very nice twython module for communicating with Twitter, so you'll need that if you don't already have it installed. Piping the tweets through sort and uniq is a suggestion rather than a requirement

./getTwitter.py <keyword> [<num pages>] | sort | uniq > twitter.txt

Then align the tweets! The cpp code is set up to read lines from a file and align each line with each other line. It writes out html to stdout, and highlighted results using ANSI escape sequences to stderr. You can redirect to suit your needs.

./nw twitter.txt > alignments.html

Mess with the code:

I've tried to comment the code well, so hopefully it is somewhat self-explanatory. src/NeedlemanWunsch.cpp contains the class that wraps the main business logic. Its usage is fairly simple:

// initialize
NeedlemanWunsch nw;
Alignment al;
nw.setStrings(A, B);
// configure (optional)
nw.setGapPenaltyFunction(someFunction);
nw.setSimilarityFunction(someFunction);

// run the algorithm
nw.align(al);

// show the results
al.printHeader(cout, Alignment::HTML);
al.print(cout, Alignment::HTML);
al.print(cerr, Alignment::CONCOLE);

The optional configuration functions allow changing the scoring scheme that the algorithm uses. setGapPenaltyFunction takes a pointer to a function that produces a penalty for a gap, based on the length of the gap. setSimilarityFunction takes two chars, and returns an int representing how similar the two chars are. Currently the similarity is based on the distance between the two characters on a standard qwerty keyboard.