T-Gene uses a gene annotation file and, optionally histone modification and gene expression data for a panel of 'tissues'. It computes a distance based score---the Distance p-value---for each potential regulatory link between one of your Loci and a transcript in the gene annotation file. Potential links are 1) all locus-transcript pairs where the transcription start site (TSS) is located within a specified maximum distance from the locus (500,000bp by default), or 2) where the transcript is the closest transcript to the locus ("Closest-TSS link"), or 3) where the locus is the closest locus to the transcript ("Closest-Locus link"). If you provide a Tissue Panel, T-Gene also computes the level of each histone modification in the panel at each of the genomic loci (typically ChIP-seq peaks) that you provide, and then it computes the Pearson correlation between a histone modification at the locus the expression of the transcript. T-Gene then estimates the p-value of this correlation, and combines it with the Distance p-value for form the CnD p-value (Correlation aNd Distance p-value). Finally, T-Gene computes the q-value of the CnD p-value, which is the minimum false discovery rate (FDR) required to consider this link statistically significant. T-Gene reports these scores for each locus-TSS-histone trio. Note that if no tissue panel is provided, T-Gene bases the q-value on the Distance p-value rather than the CnD p-value.
You provide a set of genomic loci to T-Gene via a 'locus file' in BED format. Each line in the locus file specifies a single locus specified as three fields separated by tabs: <chrom> <chromStart> <chromEnd>. Any additional fields on each line are ignored by T-Gene, as are comment lines, which must start with the '#' character.
You must choose a Genome or Tissue Panel for the exact genome release (assembly) used by your loci file for T-Gene to work correctly. If a Genome or Panel does not exist for your genome release, you may be able to use the LiftOver tool to convert your loci to the correct coordinates.
You can use T-Gene on your own computer if you download and install the MEME Suite. (See instructions in the "Download & Install" menu on the left.) You should also download the T-Gene Databases, which include all of the Genomes and Histone/Expression Panels available to the T-Gene Web Server. You can find gene annotation files for other genomes as GTF files at ftp.ensembl.org/pub and ftp.ensemblgenomes.org/pub. For example, you can download a gene annotation file in GTF format for Cow at ftp://ftp.ensembl.org/pub/release-97/gtf/bos_taurus. You can reference this file on the T-Gene command-line as long as your locus file (e.g., ChIP-seq peak file) uses Ensembl format for chromosome names and the genomic coordinates are for Ensembl release 97. (Gene annotation files for older releases are also available at the same FTP site.)
Under construction...
T-Gene allows you to select which columns to display using the check-boxes under the "Columns to display" heading on the right of the HTML output.
By default, T-Gene sorts the potential regulatory links first on the CT (Closest-TSS) column, then by Distance p-value, or by CnD p-value if you provided a tissue panel. so that links that CT links will come first. This is done because CT links are more likely to be regulatory than other links. You can change the column on which to sort the links using drop-down menu labeled "Sort", located on the right. After choosing the new sorting column, click the "Update Filter & Sort" button located below the drop-down menu.
For example, if you are interested in what the gene targets are for a transcription factor (TF), and your locus file is a ChIP-seq peak file for that TF, sorting by the default column, and leaving the filtering of links at the default (see below), makes sense. This will display the most significant Closest-TSS link for each gene at the top of the list, sorted by statistical significance, followed by the non-Closest-TSS links, also sorted by statistical signficance.
By default, T-Gene will display the best link for each gene. ("Best" depends on how you are sorting the links.) You can change this to showing the best link for each transcript ("TSSes"), or the best link for each locus ("RE Locus"), or to show all links, using the four radio buttons at the top of the "Filters" section.
In addition, you can filter on Gene ID, Gene Name, TSS ID, TSS Locus, the absolute value of the distance, the distance etc. using the check boxes and their associated fields. Note that you can use "regular expressions" to define which names and IDs you want to select with the Gene ID, Gene Name etc. filters.
Click "Update Filter & Sort" button located below the filters after changing filter selections.
You can download all the links that are displayed as a tab-separated values (TSV) file. By selecting the columns to be displayed and the filtering and sorting that you want, you can customize the downloaded file.