QC reports

[dramanica]

It would make sense to have a couple of simple functions to generate QC reports. A good place to start is plinkQC for inspiration. I would consider having 2 functions: gt_qc_loci and gt_qc_indiv. Those functions would generate each a dataframe with different columns (e.g. the loci df would have missingness, maf, hwe) and an appropriate class ('gt_qc_loci' and 'gt_qc_indiv'). We could then have a print/summary function for the two classes that creates some simple summaries, and an 'autoplot' function that generates standard plots (each 'autoplot' function could have a 'type' parameter that allows to only plot one panel or 'all' to create a composite panel.

[dramanica]

We now have 'loci_freq', 'loci_missingness' and 'loci_hwe' which should provide us with all the basic QC equivalent to PLINK.

[dramanica]

There is now a simple loci_qc_report() that returns a tibble with the appropriate class. We should test that it scales on large datasets, but otherwise it does the job. It needs a summary and autoplot method.

[eviecarter33]

Loci qc autoplots are now implemented in branch qc_plots, these need expanding with adjustable thresholds and methods for each individual plot (currently there are two autoplots - a composite plot or an 'upset' plot). I will work on qc plots for individuals next

[dramanica]

Ideally this wants its own vignette, to show off how to do qc. It would also make sense to be friendly and allow the functions to work on bigsnp objects. A visual QC is on the wish list for bigsnpr, and given how much we have benefitted from them, it would make sense to make our functions friendly to their objects.