Pangenome comparison via ED strings

This is a software suite for pangenome comparison via elastic-degenerate (ED) strings.

An ED string is a sequence of sets of strings. Our software currently supports only the DNA alphabet {A, T, C, G}, the letter N for the indeterminate base, as well as the empty string Ɛ. For example, the ED string x below has 3 sets. The first set has two strings (AGT and A); the second set has three strings (A, T, and the empty string Ɛ), and the third set has one string (ACGTN).

$ cat x.eds
{AGT,A}{A,T,}{ACGTN}

Download the source code

Using git

$ git clone https://github.com/urbanslug/junctions
$ cd junctions

or using curl and zip:

$ curl -LO https://github.com/urbanslug/junctions/archive/refs/heads/master.zip
$ unzip master.zip
$ cd junctions

Compile

Compilation is done with make and can be done in different ways.

To create a dynamically linked binary (advisable)

$ make

in case of need of statically linked binary

$ make static

When compiled with MSA support junctions is able to internally convert MSA files in RC-MSA format to ED strings however this is only supported on newer x86 processors.

To compile a dynamically linked binary

$ make WITH_MSA=true

or for a statically linked binary

$ make static WITH_MSA=true

Usage and Documentation

Run the following for the help text

$ ./bin/junctions

Further documentation can be found in the wiki.

Example 1: ED string intersection

Consider two ED strings x and y encoded in the corresponding files below:

$ cat x.eds 
{A,AC,TGCT}{CA,}

$ cat y.eds 
{,T}{GCA,AC}

We can determine whether x and y have a nonempty intersection by running the following:

$ ./bin/junctions intersect x.eds y.eds 
INFO intersection exists

Indeed, x and y share the string AC.

Example 2: ED Matching Statistics

Consider two ED strings x and y encoded in the corresponding files below:

$ cat x.eds 
{A,AC,TGCT}{CA,}

$ cat y.eds 
{,T}{GCA,AC}

We can compute their matching statistics by running the following:

$ ./bin/junctions graph -c 0 -s x.eds y.eds 
Similarity measure is: 5

In particular, the option -c 0 denotes that no constraint is imposed to the matching statistics; the option -s denotes that a similarity measure will be computed from the matching statistics.

Example 3: Breakpoint Matching Statistics

In this flavour, matching statistics are considered only between breakpoints. Consider two ED strings x and y encoded in the corresponding files below:

$ cat x.eds 
{A,AC,TGCT}{CA,}

$ cat y.eds 
{,T}{GCA,AC}

We can compute their breakpoint matching statistics by running the following:

$ ./bin/junctions graph -c 1 -s x.eds y.eds 
Similarity measure is: 3.5

In particular, the option -c 1 denotes that a constraint related to breakpoints is imposed to the matching statistics; the option -s denotes that a similarity measure will be computed from the matching statistics.

Citations

Estéban Gabory, Njagi Moses Mwaniki, Nadia Pisanti, Solon P. Pissis, Jakub Radoszewski, Michelle Sweering, Wiktor Zuba:
Comparing Elastic-Degenerate Strings: Algorithms, Lower Bounds, and Applications. CPM 2023.


Estéban Gabory, Njagi Moses Mwaniki, Nadia Pisanti, Solon P. Pissis, Jakub Radoszewski, Michelle Sweering, Wiktor Zuba:
Pangenome Comparison via ED Strings. [Front. Bioinform.](https://doi.org/10.3389/fbinf.2024.1397036)