WeaveAlign user's manual

Manual under construction
For WeaveAlign v1.2 – last updated 02 Sep 2014

Get WeaveAlign

You can choose any of two options: download a stable version (a runnable .jar file), or build your own WeaveAlign from the latest source files.
For details, visit the Downloads page.

How to use

WeaveAlign expects a set of input alignments, and will build up an alignment network of the provided alignments. Then many new alignment paths (probably different from all the inputs) will be computed and a ,,best'' one will be selected.

For a brief usage information run
java -jar WeaveAlign.jar -help

Run the example (Linux)

You can run a small example bash script:
./run-example.sh

Have a look at the script itself, it's quite simple. Here's in brief what will happen (**):
  1. 500 alignments will be unpacked from a zip file to be used as the set of input alignments
  2. The input alignments will be copied into one longer file (called a "logFile" because of its format) so that WeaveAlign can run much faster (many small files are slow to read)
  3. WeaveAlign will be called to build the network and select a good alignment
  4. The selected WeaveAlignment will be evaluated against a reference alignment, as well as the input alignments, so that we can see if we got a better result by combining the alignments (this step needs a simple config file, to be discussed later)
While the script is being run, you'll see some info printed to the screen. What you should see at the end is this:

Summary of results:
-------------------
Ref alignment size: 15 width: 144
WeaveAlign sum-of-pairs (AMA) score = 0.698
AMA scores for inputted alignment samples:
 

Percentile          5%         50%         95%
Score           0.617       0.671       0.715

Rank score = 83.6 (WeaveAlign's alignment scores better than 83.6% of the input alignments, using the sum-of-pairs score)


(**) Note: you can specify the input files and output location by editing the run_example.sh script to read in another set of .fsa alignment files, and/or modifying the config file of the evaluation step (the parameters set fot the example can be found here within the git repo: testdata/testrun500.properties).

The file that defines the parameters is really simple, just four key-value pairs, locations of input and output files:

referenceAlignment = testdata/m15-reference.fsa
weaveAlignment = testdata/wva-result-example500.fsa
dataDirOfBaseAligns = testdata/exampleInputDir/sampleAligns/
outputCsv = testrun-alignScores.csv


But be careful to edit the run_example.sh script and the properties file in a way that they play nicely together!