scikit-bio dependency

[jvollme]

when trying to run rosella i got an error during the short read processing step with flight, complaining about the missing skbio module:

[2022-09-07T12:20:35Z ERROR bird_tool_utils::command] Error when running flight process. Exitstatus was : ExitStatus(unix_wait_status(256))
[2022-09-07T12:20:35Z ERROR bird_tool_utils::command] The STDERR was: "09/07/2022 02:20:13 PM INFO: Time - 14:20:13 07-09-2022\n09/07/2022 02:20:13 PM INFO: Command - /home/ww5070/.local/bin/flight bin --assembly ../../../../assembly/220817/All_compare_normA/All_compare_normA_assembly/final.contigs.fa --input rosella_bins//rosella_coverages.tsv --kmer_frequencies rosella_bins//rosella_kmer_table.tsv --min_contig_size 1500 --min_bin_size 200000 --n_neighbors 200 --output_directory rosella_bins// --cores 8\nTraceback (most recent call last):\n  File \"/home/ww5070/.local/bin/flight\", line 8, in <module>\n    sys.exit(main())\n  File \"/home/ww5070/.local/lib/python3.8/site-packages/flight/flight.py\", line 449, in main\n    args.func(args)\n  File \"/home/ww5070/.local/lib/python3.8/site-packages/flight/flight.py\", line 565, in bin\n    rosella = rosella_engine_constructor(args)\n  File \"/home/ww5070/.local/lib/python3.8/site-packages/flight/flight.py\", line 536, in rosella_engine_constructor\n    from flight.rosella.rosella import Rosella\n  File \"/home/ww5070/.local/lib/python3.8/site-packages/flight/rosella/rosella.py\", line 48, in <module>\n    from flight.rosella.validating import Validator\n  File \"/home/ww5070/.local/lib/python3.8/site-packages/flight/rosella/validating.py\", line 49, in <module>\n    from flight.rosella.clustering import Clusterer, iterative_clustering_static, kmeans_cluster\n  File \"/home/ww5070/.local/lib/python3.8/site-packages/flight/rosella/clustering.py\", line 56, in <module>\n    from flight.rosella.binning import Binner\n  File \"/home/ww5070/.local/lib/python3.8/site-packages/flight/rosella/binning.py\", line 43, in <module>\n    import skbio.stats.composition\nModuleNotFoundError: No module named 'skbio'\n"

I did follow the installation procedure given in the readme.md, including installing flight from github using pip:
```pip install git+https://github.com/rhysnewell/flight.git#egg=flight-genome``´

I guess this means that skbio may have to be added to the dependencies? I guess this would be easiest to do by adding it to the conda recipe for rosella?

Edit: also, looking at issue #21 I guess the same applies for matplotlib, seaborn, umap, numba? Maybe it is easiest for now to just change the installation instructions to:

conda create -n rosella
conda activate rosella
conda install -c bioconda scikit-bio matplotlib seaborn umap numba rosella
pip install git+https://github.com/rhysnewell/flight.git#egg=flight-genome

[rhysnewell]

Thanks for that John, I'll fix this in future upon the next update. There was a change in some of the required dependencies and I think I might have messed up the conda recipe in the process of trying to trim back on rosella's requirements.

I'll close this once the conda recipe is updated.

Cheers,
Rhys

[jvollme]

thanks! But now after creating a new dedicated conda environment for this, i realize that the actual problem is actually with the shebang of the main script of flight (which uses the system python interpreter instead of that of the current conda environment)

[jvollme]

ok now i got it installed and running!
Shebang was fine, but there were two other problems:

the first problem (exclusively on my side) was that my conda (although updated) still installs python=2 per default, if not otherwise specified in the recipe. I solved this by explicitely specifying python3 (most more recent conda setups should not do this, but still it may be a good idea to specify python3 as a dependency in the conda recipe?)

The other problem was that the most recent version sci-kit bio in conda does not work wth the most recent version of scipy in conda. So i had to explicitely specify scipy=1.8.1

The other dependencies i thought were necessary are not actually needed to be explicitely specified.

So the updated installation procedure that worked for me is (in case anybody else has the same problem):

conda create -n rosella python=3 scipy=1.8.1 scikit-bio
activate rosella
pip install git+https://github.com/rhysnewell/flight.git#egg=flight-genome

[rhysnewell]

Thanks for working through this, I'll update the setup.py for flight, that should solve any issues