maze: match visualizer

In a similar manner to
dot plots,
maze highlights local sequence similarity between two DNA sequences. In
particular, maximal exact substring matches are computed with
MUMmer3 and visualised.

In the context of Delly this is most useful for exploring breakpoints or entire
SVs constructed with the local assembly pipeline, but maze has also been
instrumental in analysing SVs from “long read” data (PacBio, MinION) or Sanger
reads.

Our breakpoint module, which is still in an experimental state, highlights the
breakpoints of your assemblies relative to another sequence (e.g. the reference)
using local alignments computed with last-split.

Requirements

This software supports only Mac/Linux. Windows users could try our Docker image
below but this has not been tested.

For the server component, you will need Python2.7+ and the Flask and click
modules. An easy way to set this up is with a
Virtual Environment:

$ cd maze
$ virtualenv venv
$ source venv/bin/activate
$ pip install -r requirements/python2.7.txt

In addition you need to have certain binaries in your $PATH:

• mummer from MUMmer3 to compute matches
• lastdb and last-split from LAST version >= 584.

The LAST package is only needed when you plan to use the breakpoint module.

At last you will need a “modern” web browser. Chrome, Firefox and Safari should
all work. Technically, Opera and IE9+ should be fine as well although these
haven’t been tested.

Running the app

First off, start the server:

$ python maze_server.py

By default, port 5000 is used, but you can specify this with the
--port option. Now, open your web browser and head to http://localhost:5000/.

Input

Two fasta/fastq files.

• n vs. 1: multiple sequences in the query are plotted against a single
  reference sequence. Useful for reads/assemblies of a single locus. Only the
  first entry in the reference file is used.
• n vs. n: if both fasta files contain the same number of sequences, a
  pairwise comparison is triggered. Useful to scan multiple loci after
  another.

Click on the setup icon and upload the reference and query fasta files via
drag-and-drop. You can change the match type and length of MUMmer3. Hit
‘visualize’ to trigger match computation and visualization.

Usability

Use left and right keys to browse through the plots. A click on the
‘breakpoints’ button takes you to a separate page showing details on the
breakpoints.

As this software is under development we are happy about feedback via mail or
via Github.

Dockerized maze

maze is hosted on Dockerhub. To run it in Docker,
type

docker run -p 5000:5000 smei/maze

This will download the Docker image, if not yet available locally, run maze in
a Docker container and expose port 5000. Direct your browser to
127.0.0.1:5000.