To go back to TRACE home page click here.
TRACE overview
The TRACE webserver takes as input a list of query genes and a tissue, and outputs a TRACE score per query gene the reflects the likelihood that an aberration in that gene could lead to a disease that manifests in the selected tissue1.
TRACE is based on a two-layer procedure that employs multiple machine learning methods and results in tissue-specific models1. Per model, the TRACE scores of genes were precomputed using 10-fold cross validation, as explained in 1. TRACE scores can be used to prioritize the input query genes, with higher scores reflecting higher likelihood.
TRACE input
Users should upload query genes and select a tissue model.
The query genes can be uploaded in three ways:
A. A VCF file. VCF files are used in clinical settings, and can be uploaded to facilitate current analytical pipelines. For more details about the format, see https://vcftools.github.io/specs.html.
B. Common gene names. Upon typing in a gene name users need to select the correct gene from the menu.
C. A text file containing a list of query genes. Each line should contain a single common gene name.
The tissue model is selected from a menu that includes eight tissue models and two brain sub-region models. The available models are blood, brain (including separate models for cortex and cerebellum), heart, liver, muscle, nerve, skin, and testis. These models represent tissues and brain sub-regions for which at least 60 disease genes that lead to clinical phenotypes in that tissue were curated.
TRACE output
TRACE reports the TRACE scores of the query genes in the selected tissue model. The report includes a textual table and graphical representations.
TRACE scores were precomputed per tissue model, as described in 1. TRACE scores range between 0, which means that an aberration in that gene is not likely to lead to a phenotype in the tissue, and 10, which means that an aberration in that gene has a high potential to lead to a phenotype in the tissue.
The table reports the TRACE score of each query gene, along with its Ensemble Gene identifier, OMIM entry (if exists), and the rank of its TACE score relative to other query genes (a rank of 100% means top TRACE score in the input gene list).
To easily view the scores of the query genes in a different tissue model use the switch at the bottom of the output page.
The tabular view includes the following:
A. Score distribution of the query genes in the selected tissue. Note that you can hover over a point to obtain gene name and score.
B. A summary of the submitted query.
C. A gene description tab. Upon selecting a query gene from the table, RefSeq description for that gene will be presented in this tab.
References
1 Simonovksy, E. et al. Elucidating the tissue-specific impact of hereditary disease genes through interpretable machine learning models. (submitted).