Feb 2018
bpy-summarize - Generate an HTML report summarizing BAli-Phy runs.
bpy-summarize [OPTIONS] directory1 [directory2 ... ]
bpy-summarize generates an HTML report from one or more BAli-Phy run directories. The report summarizes sampled parameters, convergence diagnostics, trees, and alignments, with additional summaries depending on the model and available output. When supplying multiple run directories, use independent runs of the same analysis.
By default, the report is written to Results/index.html. Open this file in a web browser. You can generate a report while runs are still in progress and rerun the command later to update it.
Summarize one run:
bpy-summarize run-1
Combine independent runs of the same analysis:
bpy-summarize run-1 run-2
Discard the first 1000 iterations from each run as burn-in:
bpy-summarize --skip=1000 run-1 run-2
-h,
--help--skip=NUM--subsample=NUM--until=NUM--prune=TAXA--outdir=DIRECTORY--verbose--sub-partitions--node-bubblesWhen character properties were logged for a partition in every chain, the report includes posterior means, standard deviations, and medians, along with an interactive viewer in the partition's tip-alignment pages. Properties can be displayed as alignment colors to explore variation among characters. Property summaries are omitted for a partition if any chain lacks its property log.
For models that log positive-selection properties, the report includes posterior support for positive selection and ranked alignment columns, with links to the alignment viewer and complete TSV tables.
Model-averaged estimates account for uncertainty about whether positive selection is included in the model. Conditional estimates assume it is included. When selection status is fixed rather than sampled, the report states whether selection is included.
For BranchModel, the report summarizes category ω
values. For BranchModel_test, it also reports hypothesis
support and ω summaries conditional on each hypothesis. Conditional
tables include matching-sample counts but no autocorrelation or
convergence diagnostics; the ordinary parameter table provides
diagnostics for the unfiltered samples.
Under the default independent priors, an unused ω retains its prior distribution conditional on that hypothesis. Unconditional category summaries therefore include prior draws from hypotheses where the category is unused. These categories describe branch models, not variation among sites.
Paths below are relative to the output directory (Results by default).
Check commands.log in the output directory for failed subcommands.
BAli-Phy online help: https://www.bali-phy.org/docs.php.
Please send bug reports to bali-phy-users@googlegroups.com.