Dicing and loadfile generation should be about accessing the filesystem, parsing datafiles, and generating metadata such as counts files and sample sets. Presently, though, gdc_dice also generates sample heatmaps, which means it introduces a dependency upon matplotlib to generate something that is not used until the sample report is generated by gdc_report. In principle the gdc_report tool just needs to read the counts tables to re-instantiate the case_data structure that is currently used in gdc_dice to draw the heatmaps.
Dicing and loadfile generation should be about accessing the filesystem, parsing datafiles, and generating metadata such as counts files and sample sets. Presently, though, gdc_dice also generates sample heatmaps, which means it introduces a dependency upon matplotlib to generate something that is not used until the sample report is generated by gdc_report. In principle the gdc_report tool just needs to read the counts tables to re-instantiate the case_data structure that is currently used in gdc_dice to draw the heatmaps.