A draft genome scaffolder that uses multiple reference genomes in a graph-based approach.

The present document provides a short guide for using the stand-alone version of the software Medusa. This software has not yet been published. A web interface is available at http://combo.dbe.unifi.it/medusa. The source code, precompiled version and the present manual are accessible at https://github.com/combogenomics/medusa.

Medusa depends on the following packages being installed on your system and available in your PATH:

1. MUMmer: this software is available at http://mummer.sourceforge.net/.

2. Python (from 2.6) and BioPython (from 1.61).

3. Java (from 1.6).

The following Python packages should be present:

1. Networkx

2. Numpy

3. Biopython

The archive Medusa.tar.gz contains the following files:

1. A runnable .jar file medusa.jar This is the program you will run.

2. A sub-folder with python scripts needed to run the program (medusa_scripts). Leave it in the same folder of the .jar file.

3. A sub-folder with a dataset (test) that can be used to test the tool.

4. A sub-folder with scripts useful for benchmarking the tool.

The following inputs are required:

• The targetGenome file: a draft genome in fasta format. This is the genome you are interested in scaffolding.

• An arbitrary long list of auxiliaryDraft files: other draft genomes in fasta format. The closest these organisms are related to the target, the better the results will be. These files are expected to be collected in a specific directory. It is possible to specify the path to the directory, see the command "-f" in the next section.

The following output files will be produced.

• targetGenome_SUMMARY: a textual file containing information about your data. Number of scaffolds, N50 value etc..

• targetGenomeScaffold.fasta: a fasta file with the sequences grouped in scaffolds. Contigs in the same scaffolds are separated by 100 Ns by default, or a variable number of Ns (estimate of the distance between the contigs), if the option "-d" is used.

The following output files can optionally be produced.

• targetGenome_distanceTable: a tabular file with the estimation of the distance between successive contigs (bp).

• targetGenome_network.gexf: the contig network in gexf format.

• targetGenome_cover.gexf: the final path cover in gexf format.

The project folder must contain:

• the targetGenome in fasta format.

• the medusa.jar file

• the scripts sub-folder “medusa_scripts”.

• the comparison genomes sub-folder “drafts”. (In alternative you can specify another path for this folder usinf the "-f" option)

Medusa can be run with the following parameters:

1. The option -i is required and indicates the name of the target genome file.

2. The option -o is optional and indicates the name of output fasta file.

3. The option -v (recommended) print on console the information given by the package MUMmer. This option is strongly suggested to understand if MUMmer is not running properly.

4. The option -f is optional and indicates the path to the comparison drafts folder.

5. The option -random is available (not required). This option allows the user to run a given number of cleaning rounds and keep the best solution. Since the variability is small, 5 rounds are usually sufficient to find the best score.

6. The option -w2 is optional and allows for a sequence similarity based weighting scheme. Using a different weighting scheme may lead to better results.

7. The option -d allows for the estimation of the distance between pairs of contigs based on the reference genome(s): in this case the scaffolded contigs will be separated by a number of N characters equal to this estimate. The estimated distances are also saved in the "*_distanceTable" file. By default the scaffolded contigs are separated by 100 Ns.

8. The -gexf is optional. With this option the gexf format of the contig network and the path cover are porvided.

9. The option -n50 allows the calculation of the N50 statistic on a FASTA file. In this case the usage is the following: java -jar medusa.jar -n50 <name_of_the_fasta> All the other options will be ignored.

10. Finally the -h option provides a small recap of the previous ones.

When medusa archive is unzipped the following files will be extracted:

• the medusa.jar file.

• the scripts sub-folder “medusa_scripts”.

• the utility test scripts folder "medusa_testing"

• a folder “test”, containing one test bacterial datasets.

java -jar medusa.jar -f test/reference_genomes/ -i test/Rhodobacter_target.fna -v

Additional datasets can be retrieved at the medusa_datasets repository https://github.com/combogenomics/medusa_datasets.

Just type

git clone https://github.com/combogenomics/medusa_datasets.git

The project can be compiled by calling ant in the top-level directory:

ant