pbradleylab / phylogenize Goto Github PK

This indicates that the BURST binary couldn't be found. Check that you have the correct binary for your operating system and architecture, that it is installed to wherever burst_dir is set to point, and that you can run the BURST binary by itself.

Solution to error: Fontconfig error: unable to match font pattern on macOS

If you see the error Fontconfig error: unable to match font pattern, you may need to install or reinstall some packages using Homebrew. From the command line, after installing Homebrew:

brew update
brew upgrade
brew reinstall cairo

Then, open R and in a fresh session, run:

devtools::install_github("davidgohel/gdtools") 
devtools::install_github("davidgohel/rvg")
devtools::install_github("davidgohel/ggiraph")

Solution to: problems with phytools::fastAnc
First check you R version and make sure it is newer than 4.0. If that doesn't work there is an important liamrevell/phytools#47 (thanks to Liam Revell and Guangchuang Yu for fixing this so quickly) you should make sure you are using.

This means that you have too few taxa mapped to MIDAS IDs in every phylum phylogenize tests. "Too few" here is defined by the options treemin (minimum taxa per phylum as an absolute number, default: 5) and pctmin (minimum taxa per phylum as a fraction of all taxa in that phylum, default: 0.01, range: 0 to 1.0). You can try lowering the cutoffs, but they are in place because with so few observations, even if you were to get gene hits with phylogenize, the results probably would not be that meaningful.

This situation can happen if you, for example, are using a "testing" dataset with very few ASVs, if your read depth was extremely low and few ASVs were detected, or if you are sequencing a community where almost nothing maps to the MIDAS database. If you are using a testing dataset, using the full dataset should solve the problem (if you really want to have something fast to run, you could try only keeping ASVs that map to a single phylum). Assuming you started with lots of ASVs, taking a look at the intermediate BURST output file output_assignments.txt should tell you how many of them were successfully mapped to MIDAS IDs: if there are only a few entries with a percent identity above 98.5% (default), lack of reference genomes is likely your problem.

Update (11/4/2019): in the latest version of phylogenize you should get a more informative error message; please file a new issue if this still happens!

There are a couple of things to try:

• If your conda/QIIME2 installation is not brand new, try removing your existing installation and installing miniconda3 again from scratch. This has helped me before, possibly because there was stuff installed in the 'base' conda environment that was conflicting with requirements for phylogenize.
• Try using the mamba environment solver, instead of the one built into conda.
• Try using strict channel priority.