How to generate the analysis/figures from “More efficient drugs lead to harder sweeps of drug resistance in HIV-1”

To be run once:

align.r

This script cleans the data and generates various parsed and clean versions of the data that are stored in /tmp and will be used by other scripts. After align.r is run once, everything else requires only read.in.data.r besides other dependencies listed explicitly.

To do the validations for the ambiguous reads, run in order:

ambread-validation-prep.r
ambread-validation-graph.r

Note: the graph from this figure isn’t actually in the paper (only the correlation coefficient).

Producing figures:

To produce the following figures that don’t require the fitting of GLMMs:

• Figure 2 (F2, F2-S1, F2-S2)
• Figure 5 (F5-S1, F5-S2)

make-nonmodel-figs.r

The rest of the figures in Figure 5 (F5-S3, F5-S4, F5-S5) are produced in the top of subsample-run.r. subsample-run.r also fits and saves all of the models that are eventually plotted in F3 and F4. It fits subsampled GLMs and GLMMs for three different iterations of the data:

1. All data, p-thinning to the 1995 level
2. Data truncated to 4 DRMs, p-thinning to the 1989 level
3. Data truncated to 4 DRMs, p-thinning to the 1995 level

Most of subsample-run.r runs slowly (i.e., days when parallelized to four cores on a 2016 macbook pro). I’ve therefore lowered the iterations from the 1000 shown in the text to 20, which should give you an idea without running for days.

Versions of Figure 3 and Figure 4 for each of data iterations 1, 2 and 3 above can be plotted with subsample-plot-etc.r.