Learn Peak Shape Filter

The Learn Peak Shape Filter tool allows you to build a new peak shape filter from sequencing data and a set of positive and negative regions. The resulting filter can be used to identify genomic regions whose read coverage profile matches the characteristic shape of the positive examples and does not match the shape of the negative examples. The procedure used to build the peak shape filter is described here Learning peak shapes).

An example of such filter is shown in figure 36.30.

Image filter2020_lowq
Figure 36.30: Example of a peak shape filter with a window size of 400bp made up of 20 bins of size 20bp each. The filter was built from ChIP-Seq data of the transcription factor NRFS and a control ChIP-Seq experiment.

To run the Learn Peak Shape Filter tool:

        Toolbox | Epigenomics Analysis (Image epigenomics) | Advanced Peak Shape Tools (Image adv_epigenomics) | Learn Peak Shape Filter (Image learn_peak_shape_tools)

This will open up the wizard shown in figure 36.31 where you can select the input data (for example the mapped ChIP-Seq reads). Track based read mappings (Image read_track_16_n_p) and stand-alone read mappings (Image contig) / (Image multicontig) are both accepted. Multiple inputs are accepted, provided that they refer to the same genome.

Image PQS_learn_wiz_step1
Figure 36.31: Select the input data for the Learn Peak Shape Filter tool.

Click Next to go to the next wizard step (shown in figure 36.32).

Image PQS_learn_wiz_step2
Figure 36.32: Options for Learn Peak Shape Filter.

In this wizard step you have the following options:

The result of the algorithm will be a Peak shape filter (Image graph), which can then be applied to call peaks or score regions using Apply Peak Shape Filter. After clicking on the button labeled Next, you can choose whether you want to open the result directly, or save the results in the Navigation Area. If you choose to save the results, you will be asked to specify where you would like to save them.