Miniconda下scikit-bio与Python 3.5兼容冲突问题求助

Hey there, I’ve run into similar conda dependency headaches before—let’s get this sorted out for you! The core issue here is that the latest version of scikit-bio available on conda requires Python 3.6+, but you’re trying to use Python 3.5 for your dsfdr2 setup. Luckily, there are a few straightforward fixes:

Solution 1: Specify an older, Python 3.5-compatible scikit-bio version in your conda create command #

The PyPI info you referenced notes scikit-bio 0.2.3 works with Python 3.4+, and this version is available on conda. Just modify your original command to pin scikit-bio to 0.2.3:

conda create -n dsfdr2 python=3.5 numpy scipy jupyter scikit-learn scikit-bio=0.2.3 statsmodels

This tells conda to pull the exact version of scikit-bio that’s compatible with Python 3.5, avoiding the dependency conflict entirely.

Solution 2: Create the environment first, then install scikit-bio via pip #

If conda still throws issues with the pinned version, you can build the base environment without scikit-bio, then use pip to install the compatible version directly:

1. Create the environment without scikit-bio:

   conda create -n dsfdr2 python=3.5 numpy scipy jupyter scikit-learn statsmodels
   

2. Activate the environment:

   conda activate dsfdr2
   

3. Install the Python 3.5-compatible scikit-bio version using pip:

   pip install scikit-bio==0.2.3
   

This bypasses conda’s strict dependency checks and lets you install the exact version you need.

Solution 3: Upgrade Python to 3.6 (if dsfdr2 supports it) #

If your dsfdr2 code doesn’t strictly require Python 3.5, you can switch to Python 3.6 to use the latest scikit-bio version without conflicts. Run this command instead:

conda create -n dsfdr2 python=3.6 numpy scipy jupyter scikit-learn scikit-bio statsmodels

Just double-check first that the dsfdr2 tool works correctly with Python 3.6—if it does, this is the cleanest long-term fix.

Quick Optional Tip #

If you run into slow download speeds on Windows, you can add a domestic conda mirror to speed up package pulls, but that’s only necessary if you’re having trouble fetching dependencies.

内容的提问来源于stack exchange，提问作者pmaes