SjTree

SjTree is a tool to analyze variance stabilizing transformation (VST) normalized expression value of RNAseq. It is build around a composite model based on xgboosted trees. The composition of this model may change based on future work, but the current implementation will still be available, along with all of these possible iterations. Data interpolation work is still ongoing and we are currently evaluating different strategies to enable deployment of this tool on different datasets.

Script usage

System requirements

SjTree runs on the Linux distribution (tested on Ubuntu 20.04 LTS) and requires python 3.7. + With the packages listed in the requirements.txt file. We recommend hardware with at least 2+ GB of RAM for this program for minimum performance and 12+ GB for optimum performance.

Installation guide

The installation procedure (including installing the required dependencies) should take less than 5 minutes with an internet speed of 5 Mbps or more.

Clone the current directory on your computer

git clone https://github.com/LBAI-InfoLab/SjTree.git

Go the downloaded directory

cd SjTree/

(optional) install SjTree as a python module

python setup.py install

Install the required dependencies

pip install -r requirements.txt

Usage

Navigate to the program directory, you should see a SjTree.py file that you can use as a python script.

python SjTree.py -h

Arguments

• --ifile / -i : the input file, dataset to process
• --ofile / -o : name of the output file to generate
• --interp / -p : set to False to skip interpolation (optionnal)
• --help / -h : display help

Ouput

In addition to the predictions (stored in the user-defined output file), SjTree create a cluster distribution figure (.png file) and an AFD representation of clusters (.gif) in the images subfolder.

Example

To run SjTree on your own data, use the following line:

python SjTree.py --ifile my_data_file.csv --ofile my_result_file.csv

Demo

The "example" folder contains a file with anonymized data from the original study and the associated expected results. You can test the software using the following command line:

python SjTree.py --ifile example/example_dataset.csv --ofile predictions.csv --interp False

You can then check that the predictions correspond to the expected results (available in the example folder). The script should take approximately 40 seconds to run on a computer with the recommended configuration.

Data structure

Mandatory columns

The input file must be a csv file containing at least the following columns:

• ID
• NAPA
• RSPH9
• TMEM92
• OAS1
• MXD3
• USP18
• ILK
• GTPBP2
• PI4K2B
• RP11.295G20.2
• ZNF318
• IFI27
• MMP25
• STIL
• MX1
• LINC01001
• NCF4
• WDR89
• LIMK2
• CCT2
• ZCCHC2
• HERC5
• DHX58
• USP32
• PGS1
• ANXA3
• OASL
• BCORL1
• IFI44L
• TLR1
• KCNJ2
• CHST15
• HAL
• YKT6
• CLUAP1
• TBL1X
• CACUL1
• KIAA1045
• THADA
• ETS2

Model composition

C4 Detector

C1C2C3 Classifier

C1

C2

C3

Citing Sjtree

Please cite this programm using the following reference:

Foulquier N, Le Dantec C, Soret P, A new molecular classification to drive precision treatment strategies in primary Sjögren's syndrome, SjTree, DOI: 10.5281/zenodo.4643639, 2020