Table of Contents Perform statistical significance test of tandem repeats. Read in tandem repeat annotations. Perform a statistical significance test. Filter tandem repeat below a significance threshold. Available significance tests Read in tandem repeat annotations. Perform a statistical significance test. Filter tandem repeat below a significance threshold. Available significance tests Annotate tandem repeats from sequence domain models. Perform overlap filtering of redundant tandem repeat annotations. Here you learn how to perform statistical significance tests on your candidate set of tandem repeats. TRAL currently provides a variety of model-based and none-model based tests. They are detailed in a NAR publication (2012) . Requirements for this tutorial: Install TRAL . TRAL ships with the sequence data needed for this tutorial. Install TRAL . TRAL ships with the sequence data needed for this tutorial. Download p-Value distribution files . Download p-Value distribution files . Read in tandem repeat annotations. ¶ There are nine proteins in the HIV proteome: Perform a statistical significance test. ¶ Calculate the statistical score on the ‘denovo’ repeat_list , and the corresponding p-Value with parameters defined in the configuration file : For example, the following putative tandem repeat is found to be non-significant with the used model of tandem repeat evolution (interpretation) :

Whereas this putative tandem repeat is considered significant: Besides, parameters can also be directly supplied to calculate_pvalues : Filter tandem repeat below a significance threshold. ¶ The resulting result_list now only contains tandem repeats with a p-Value below 0.05, and is shorter than the original tandem repeat list: Available significance tests ¶ The model based scores are generally described in our NAR publication, 2012 . Briefly, the sequences evolution is assumed to be described by the LG matrix. Further, the models can be distinguished by how gaps are handled: ‘phylo’: Gaps are treated as characters. ‘phylo’: Gaps are treated as characters. Gaps can also assumed to be inserted with exponentially distributed waiting times and Zipfian distributed indel lengths: ‘phylo_gap01’, ‘phylo_gap01_ignore_trailing_gaps’, ‘phylo_gap01_ignore_coherent_deletions’, ‘phylo_gap01_ignore_trailing_gaps_and_coherent_deletions’: Gaps are modelled, and assumed to mutation at at 0.1 times lower rate compared to substitutions. ‘phylo_gap01’, ‘phylo_gap01_ignore_trailing_gaps’, ‘phylo_gap01_ignore_coherent_deletions’, ‘phylo_gap01_ignore_trailing_gaps_and_coherent_deletions’: Gaps are modelled, and assumed to mutation at at 0.1 times lower rate compared to substitutions. ‘phylo_gap001’, ‘phylo_gap001_ignore_trailing_gaps’, ‘phylo_gap001_ignore_coherent_deletions’, ‘phylo_gap001_ignore_trailing_gaps_and_coherent_deletions’: Gaps are modelled, and assumed to mutation at at 0.01 times lower rate compared to substitutions. ‘phylo_gap001’, ‘phylo_gap001_ignore_trailing_gaps’, ‘phylo_gap001_ignore_coherent_deletions’, ‘phylo_gap001_ignore_trailing_gaps_and_coherent_deletions’: Gaps are modelled, and assumed to mutation at at 0.01 times lower rate compared to substitutions. ignore_trailing_gaps signifies that gaps before the first tandem repeat unit and after the last tandem repeat unit are not penalised. ignore_trailing_gaps signifies that gaps before the first tandem repeat unit and after the last tandem repeat unit are not penalised. coherent_deletions signifies that gaps of the same length and position within the alignment of tandem repeat units are penalised only once. coherent_deletions signifies that gaps of the same length and position within the alignment of tandem repeat units are penalised only once. Among the three ad hoc scores are ‘entropy’, ‘parsimony’, and ‘pSim’ scores (see our NAR publication, 2012 ).