A tool to model the evolution and structural impact of alternative splicing

Status	Linux, OSX	Windows

PhyloSofS (Phylogenies of Splicing isoforms Structures) is a fully automated computational tool that infers plausible evolutionary scenarios explaining a set of transcripts observed in several species and models the three-dimensional structures of the produced protein isoforms.
The phylogenetic reconstruction algorithm relies on a combinatorial approach and the maximum parsimony principle. The generation of the isoforms' 3D models is performed using comparative modeling.

PhyloSofS was applied to the c-Jun N-terminal kinase (JNK) family (60 transcripts in 7 species). It enabled to date the appearance of an alternative splicing event (ASE) resulting in substrate affinity modulation in the ancestor common to mammals, amphibians and fishes, and to identify key residues responsible for such modulation. It also highlighted a new ASE inducing a large deletion, yet conserved across several species. The resulting isoform is stable in solution and could play a role in the cell. More details about this case study, together with the algorithm description, can be found in the PhyloSofS' preprint available at bioRxiv.

You can clone this PhyloSofS package using git:

git clone https://github.com/PhyloSofS-Team/PhyloSofS.git

Then, you can access the cloned PhyloSofS folder and install the package using Python 3's pip:

cd PhyloSofS
python -m pip install .

To run the phylogenetic module of PhyloSofS, you need to have Graphviz installed.
The easiest way to install Graphviz in...

• Debian/Ubuntu is: sudo apt-get install graphviz
• Windows is using Chocolatey: choco install graphviz
• macOS is using Homebrew: brew install graphviz

The molecular modelling pipeline depends on Julia, HH-suite3 and MODELLER. This module can only run on Unix systems (because of the HH-suite). To alleviate that, we offer a Docker image with all these dependencies installed (see the Docker section for more details).

You can download Julia 1.1.1 binaries from its site.

Some BioJulia packages can need LibZ to precompile. If you found a related error, you can install LibZ from its site. In Ubuntu 18.04 you can install it by doing: sudo apt-get install zlib1g-dev

Clone our HH-suite fork at AntoineLabeeuw/hh-suite and follow the Compilation instructions in its README.md file.

PhyloSofS needs MODELLER version 9.21. Follow the instructions in the MODELLER site to install it and get the license key.

To run the molecular modelling module you need the HH-suite databases:

• Sequence database: uniclust30_yyyy_mm_hhsuite.tar.gz (we have tested PhyloSofS using 20180_08 as yyyy_mm)
• Structural database: pdb70_from_mmcif_latest.tar.gz

The needed mmCIF PDB files for MODELLER are downloaded on demand, if there are not present, in an indicated folder.

To set up the databases, you can use the script setup_databases (recommended). Alternatively, a manual installation can be performed following the instructions in docs/get_databases.md.

The setup_databases downloads and decompress the needed databases. It creates the following folder structure that can be easily used by PhyloSofS with the --databases argument:

databases
 ├── pdb
 ├── pdb70
 └── uniclust

You can do setup_databases -h to know more about the script and its arguments.

You can directly use PhyloSofS via Docker without cloning this GitHub repository. To run PhyloSofS' Docker image you need to install Docker following these instructions.

The following example is going to run PhyloSofS' Docker image using Windows PowerShell. Databases for the molecular modelling module stored in D:\databases are going to be mounted in /databases and the local directory in /project. The actual folder is ${PWD} in Windows PowerShell, %cd% in Windows Command Line (cmd), and $(pwd) in Unix.

docker run -ti --rm --mount type=bind,source=d:\databases,target=/databases --mount type=bind,source=${PWD},target=/project diegozea/phylosofs

After this, we have access to the bash terminal of an Ubuntu 18.04 image with PhyloSofS and all its dependencies installed. You only need to indicate your MODELLER license key to use PhyloSofS. To do that, you run the following command after replacing license_key with your MODELLER license key:

sed -i 's/xxx/license_key/' /usr/lib/modeller9.21/modlib/modeller/config.py

After installing Docker CE following these instructions, you can create a folder to work with the app, e.g.:

mkdir phylosofs

And then go into that folder and run the PhyloSofS Docker image bind-mounting the local folder into /project:

cd phylosofs
sudo docker run -ti --rm --mount type=bind,source=$(pwd),target=/project diegozea/phylosofs

This starts a bash console with PhyloSofS and all its dependencies installed. The sources are taken from diegozea/phylosofs. First, change xxx by your MODELLER license key using sed as indicated in the banner. Then, you can use the setup_databases script the first time to install the needed databases into the project folder. The databases are going to need some time to download and decompress depending on your internet connection and disk speed. You need almost 129 Gb in your disk before download and decompress them:

setup_databases

This has created a databases folder in /project (and therefore in the phylosofs folder of your system) with the needed sequence and structure databases for the homology modelling step.

To test the molecular modelling suite, we are going to create an example input pir file in a GeneName folder. PhyloSofS is going to look for transcripts.pir files in the indicated folder and its subfolders:

mkdir GeneName
echo ">P1;gene transcript ABCDE" >> ./GeneName/transcripts.pir
echo "AAAAAABBBBBBBBBBBBBBBBBBBCCCCCCCCCCCCCCCCCDDDDDDDDDEEEEEEEEE" >> ./GeneName/transcripts.pir
echo "ACTNEFCASCPTFLRMDNNAAAIKALELYKINAKLCDPHPSKKGASSAYLENSKGAPNNS*" >> ./GeneName/transcripts.pir

Where the pir annotation below the id is used to indicate the exon (A, B, C...) to which belong each residue of the protein isoform.

phylosofs -M -i GeneName --databases databases

If you are using the PhyloSofS' Docker image, you must know that errors can occur when very large files are being written to bind-mounted NTFS file systems. This happens particularly when setup_databases is run because it tries to download and decompress large files. To avoid this problem, you can install PhyloSofS on Windows and run setup_databases.exe to set up the databases before using the docker image.

You can run phylosofs -h to see the help and the list of arguments.

phylosofs -P -s 100 --tree path_to_newick_tree --transcripts path_to_transcripts

If databases where installed using setup_databases and the HH-Suite3 scripts and programs are in the executable paths, then you can run:

phylosofs -M -i path_to_input_dir --databases path_to_databases_folder

PhyloSofS is going to look for transcripts.pir files in the folder and sub-folders of path_to_input_dir to perform the homology modelling of each sequence in those files.

If you have a more manual installation of the databases and/or the HH-Suite3 scripts and programs are not in the path:

phylosofs -M -i path_to_input_dir --hhlib path_to_hhsuite_folder --hhdb path_to_uniclust_database/uniclust_basename --structdb path_to_pdb70/pdb70 --allpdb path_mmcif_pdb_cache_folder

Please note that for the databases --hhdb and --structdb, you need to provide the path to the folder and also the basename of the files in it. For example, if the database uniclust30_2018_08 is located in /home, you need to write:

--hhdb /home/uniclust30_2018_08/uniclust30_2018_08

You can also find useful the arguments:

• --ncpu number_of_cpu
• --julia path_to_julia_executable

If you installed the databases using setup_databases in a folder that you have bind-mounted to /databases, then you only need to run:

phylosofs -M --databases /databases

Because the Docker image has HH-Suite3 installed with its programs and scripts in the executable paths.

The PhyloSofS package has been developed under the MIT License.

For questions, comments or suggestions feel free to contact Elodie Laine or Hugues Richard