multiVitamin

In-Terminal multiple alignment tool for graphs

multiVitamin is a software package which allows users to perform multiple alignments with graphs. There are three algorithms available for this task: The Bron-Kerbosch and the VF2 algorithm and a custom subgraph-isomorphism-VF2 (called subVF2). Details on the algorithms can be found in the theory section here. The main function of the package is to align two or more graphs following a binary alignment guide tree containing the "best" alignment procedure. The resulting multiple graph alignment is represented as one single graph itself. Next to this main function, there is also the possibility to show all co-optimals of an alignment between 2 graphs or to view graphical representations of graphs. Sounds exciting, so let's get started!

Comments for practical course

Changes from 18.02

• If you want to create an interactive representation of a graph without (re)running an alignment, simply run multiVitamin -rV <graph file>. The html file will be generated in your results directory without opening a popup (in order to facilitate automization)

Changes from 07.02 to 11.02

• Important: Please reinstall multiVitamin after pulling.
• added -r, a new dynamic, interactive graph representation of the consensus graph (example: multiVitamin -ri <g1.graph> <g2.graph> ). The script takes the string which appears most often in the label list. The nodes are colour coded: multiVitamin recognizes elements and atomic numbers and gives them appropriate colours. If an element occurs more often in a node label, the node's colour gets more dark. If an element is missing (i.e. it is displayed as a black node), simply add it in multivitamin/supp/molecule_dicts.py. Double/triple/... bonds are not supported as of now. Hovering over a node displays its original label list
• Added a progress bar to subVF2 that gives (little) indication on the estimated progress (it counts how many times VF2 has returned to 0 depth, that is, has rechosen the first two mapped nodes).
• added support for custom guide_trees. Simply write a file which contains the guide tree in the first line in Newick format. Example: (((g1,g2),(g3,g4)),g5); Then use the -m flag and indicate your file path and name. The program will parse the file and show you a graphical representation of the guide tree in order for you to verify its structure.

Please report any bugs to me, so that I can fix them asap without you losing too much time.

Changes from 04.02 to 05.02

• added the -o flag with which you can specify the name of your resulting multiple alignment graph.
• fixed the very ugly node IDs in the output_graphs. They now look better (relatively) and are actually reparsable, i.e. realignable. Nonetheless, you might want to use the graph with shorter IDs generated with -s if you are realigning it.
• The scoring table layout changed, because there were issues with some editors concerning the tab spaces. This is why tabs were replaced by labelsep which you can specify in custom.py. The default is space. If you use labels that include spaces, you should change labelsep accordingly.
• added some more comments to the functions and the code in general here and there.

Please note

• At the beginning of your day, please take the time to pull and reinstall multiVitamin (if you previously installed it via pip3). For the rest of the day, unless otherwise communicated, pulls from time to time should be enough, you should not need to reinstall the package after each pull.
• Due to time reasons, many useful features are not indicated/explained in the multivitaminReadme.pdf, if you experience any trouble with the program, feel free to ask. I will try and add the missing information to the Readme as soon as possible.
• The subVF2 algorithm is the only one suitable for multiple alignment. You do not need to specify it as it is also the default algorithm.
• The greedy multiple alignment algorithm is the only one available at the moment due to its acceptable running time. If you experience otherwise, please tell me and I will implement a progressive version.
• There are many useful features in the custom.py script explained inside.
• Many changes had to be done last minute, unfortunately. So you could quite likely experience some bugs here and there. Please let me know, if anything appears odd to you. Many excuses in advance!

Installation instructions

Clone the repo from github with

git clone https://github.com/mk36fyvy/multivitamin.git

There are two possibilities to run multiVitamin.

pip3 install

Navigate to the directory containing the setup.py file

cd multivitamin/

If you have root permissions, run

pip3 install -e .

and you're good to go.

If you do not have root permissions, run

pip3 install -e . --user

to install multiVitamin locally. Then put it into your PATH, so it can be run from everywhere within the system.

echo 'export PATH=$PATH:~/.local/bin/' >> ~/.bashrc

After this, you will need to either run source ~/.bashrc or open a new terminal window.

Done, now you can run multiVitamin in the command shell. You can test if the installation was successful by typing

multiVitamin -h

Without installation

You can also run the package without installation by running

[path_to_dir]/multivitamin/run.sh 

followed by the standard input specified in

[path_to_dir]/multivitamin/run.sh -h 

How to use

Basics

You can get an overview of the basic flags and functionalities by typing

multiVitamin -h

A basic run example looks like the following

multiVitamin -a subVF2 -i graph1.graph graph2.graph graph3.graph

or simply

multiVitamin -i .

This will align all the .graph files (mind the extension!) in the current working directory.

Examples on what our custom graph file format looks like can be found in graph_examples inside the module.

Custom scoring

Pairwise alignments are scored in a very generic and basic way by default: exact matches are rewarded, gaps are punished, both with fixed values (4 and -1 respectively, editable here in score_without_matrix()).

With the flag -t you can provide a scoring table with custom sum-of-pair scores. A table example can be found here.

If you do not want to type all the pairings of the labels yourself, just specify all single labels like so

 -\t-2
 C\t2
 N\t2

where \t indicates a tab spacing.

This specifies the exact match score, i.e for instance match(C,C) (the gap symbol - has to be indicated in the first line, if you want a different gap score than -1. Simply omit it, if this is not the case.).

If you run the program with this minimal scoring table, you will get a warning with all the missing pairs written out. Like that, you can simply copy and paste those of interest into your scoring table file, change the scores and rerun the program.

You can also forbid the alignment of nodes with specific labels. Check custom.py for this.

Further reading

More information on how to use the module can be found here.