TiSAn:

This folder contains code to run TiSAn-build, the companion app for tissue-specific feature extraction.

Panel1: expression Quantitative Loci (eQTL)

In this panel, the user needs to provide a database of known tissue-specific eQTLs (e.g., GTEx consortium).
Once a directory has been selected, the program will go through all the different tissues available in the database.
A checkbox list will appear, and the user can select all the tissue types he considers as 'of interest' for a set of positive annotations.
By defaut, the remaining tissues will be held as 'non-tissue' eQTLs.
Once the tissue(s) of interest are selected in the checkbox list, the user can click on the 'eQTL tissue selection done' button.
After a couple of minutes, two eQTL databases (tissue ('tissue_eqtl_gr.Rdata') and non-tissue ('non_tissue_eqtl_gr.Rdata')) should appear in the working directory.
NB: if a file named 'tissue_eqtl_gr.Rdata' or 'non_tissue_eqtl_gr.Rdata' already exists, the program will not overwrite it.

In this panel, the user needs to provide a database of known tissue-specific methylation profiles (e.g., RoadMap Epigenomics consortium).
The program expects one single file containing all the methylation levels measured in different cell lines at different loci, and a metadata file contianing the information regarding the different sampled tissues.
Once these two files have been provided, the program will go through all the different tissues available in the database.
A checkbox list will appear, and the user can select all the tissue types he considers as 'of interest' for a set of positive annotations.
By defaut, the remaining tissues will be held as 'non-tissue' DMRs.
Once the tissue(s) of interest are selected in the checkbox list, the user can click on the 'Methylation tissue selection done' button.
After a couple of minutes, a DMR database ('rme_gr.Rdata') should appear in the working directory.
Here, it contains, for thousands of loci, the average DNA methylation level found in tissue and non-tissue cell lines.
NB: if a file named 'rme_gr.Rdata' already exists, the program will not overwrite it.

In this panel, the user needs to provide a database of known genes in their corresponding publications (e.g., gene2pubmed).
The program expects one single file containing the pairs of genes and publications.
The program also expects key words related to the tissue of interest (e.g., 'brain').
Once it has been provided, the user can click on the 'tissue-specific keywords provided' button and the program will go through all the publications and search for keywords (it might take half an hour).
Once the tissue-related genes are identified, they will be stored in 'tissue_gene_gr.Rdata'.
The remaining genes will be stored in 'other_gene_gr.Rdata'.
NB: if a file named 'tissue_gene_gr.Rdata' or 'other_gene_gr.Rdata' already exists, the program will not overwrite it.

In this panel, the user needs to provide a database of known disease-associated long intergenic non-coding RNAs (lincRNA) (e.g., LincSNP).
The program expects one directory with one file per chromosome containing the pairs of SNPs and diseases.
Once this directory has been provided, the program will go through all the different diseases available in the database.
A checkbox list will appear, and the user can select all the tissue-related diseases he considers as 'of interest' for a set of positive annotations.
Once it has been provided, the user can click on the 'lincSNP disease selection done' button and the program will go through all the known associations (might take 5 minutes).
Once the disease-tissue SNPs are identified, they will be stored in 'train_lincsnp_pos.txt'.
The remaining loci will be stored in 'train_lincsnp_neg.txt'.
NB: if a file named 'train_lincsnp_pos.txt' or 'train_lincsnp_neg.txt' already exists, the program will not overwrite it.