bio-miga/miga

Scythe no longer in conda

lmrodriguezr opened this issue · 10 comments

lmrodriguezr commented

Originally posted by @BJCampbelllab in #124 (comment)

Hi Luis,
I ended up removing the miga environment from conda by rm the environment/directory and am attempting to reinstall from these instructions:
https://manual.microbial-genomes.org/part2/requirements/conda

I got an error for scythe. It was no longer found in faircloth-lab.

PackagesNotFoundError: The following packages are not available from current channels:

  • scythe

Current channels:

I went to the anaconda page and searched for scythe, and that only gave me access to a scythe-score-counter (tpug).

Any suggestions?
Thanks!
Barb

lmrodriguezr commented

Hello @BJCampbelllab ,

We're working on replacing scythe altogether, so this problem will soon disappear. However, in the meantime, I'd like to ask you: Do you want to process raw reads or only assembled genomes?

BJCampbelllab commented
lmrodriguezr commented

Perfect. Since you don't need read processing, my recommendation would be:

  1. Update MiGA to the latest version (v0.7.26.0)
  2. Initialize miga again using: miga init --no-optional (or if you want MyTaxa and RDP on, simply: miga --no-read-processing)
  3. Now MiGA should work without problems as long as the input is assemblies

Also, when running the workflow, I'd strongly recommend using the flags --fast and --haai-p fastaai. These will become the default in the upcoming v1.0.

Related issues: #107 and #99

BJCampbelllab commented
lmrodriguezr commented

Make sure you have all the rest of the prerequisites (excluding those for read processing):

conda install -y -c conda-forge r-base r
conda install -y --strict-channel-priority -c conda-forge ruby
conda install -y sqlite openjdk
conda install -y -c conda-forge -c bioconda \
  blast hmmer bedtools prodigal mcl barrnap \
  diamond krona fastani

And then just rerun:

gem install miga-base
miga init --no-optional

FastANI should be installed in the conda step and detected by miga init in the last step. The MiGA version should be 0.7.26.2 (there was an issue with the GitHub actions that prevented v0.7.26.0 to ship with ancillary software, including FastAAI).

Best,
Miguel.

BJCampbelllab commented
lmrodriguezr commented

Great!

To see the index.html, you could transfer that file as well as the browse folder. However, that's only there to help you navigate the files (and to show some of the summary statistics).

Yes, you are correct: the data you need is stored in data/09.distances/02.aai/miga-project.txt.gz. The format is a little strange, but you can easily transform it into a regular matrix in R using the enveomics.R package (which MiGA should have installed already):

library('enveomics.R')
a <- read.table(gzfile('data/09.distances/02.aai/miga-project.txt.gz'), head = TRUE)

# Square AAI similarity matrix:
sim.aai <- as.matrix(enve.df2dist(a, 'a', 'b', 'value'))

# Or directly as a distance matrix:
dist.aai <- enve.df2dist(a, 'a', 'b', 'value', max.sim = 100)

Best,
Miguel.

BJCampbelllab commented
lmrodriguezr commented

Dear Barb,
I'll close this issue now, but please feel free to reopen if you encounter any new problems (or if you continue to experience this one).

Best,
Miguel.

BJCampbelllab commented