Maintain the tutorial

[alneberg]

The bam mapping step is really old and seems to cause several issues. Perhaps change to just using mapping with some regular mapping and skip the markduplicates step.

• [ ] Use CheckM instead of our old custom SCGs
• [ ] Fix broken links in tutorial
• [ ] Go through Docker instructions
• [ ] Use Megahit instead of velvet
• [ ] Remove evaluation except for CheckM
• [ ] Remove linkage
• [ ] Add script that can plot checkm results
• [ ] Possibly some other evaluation?

Next iteration:

• [ ] Use bwa mem instead of our current bash script
• [ ] Use samtools idxstats instead of coverage bed
• [ ] Update gen_input_table.py to be able to use idxstats input

[franciscozorrilla]

Really looking forward to these upcoming changes, any idea when they will be implemented?

[alneberg]

Thanks for the support, @franciscozorrilla!

I can't really give a time estimate of this since I am working on it only a few hours per week due to other work duties... :(

[franciscozorrilla]

No problem! I understand. Also, I wanted to ask why the linkage stuff will be removed in the future?

[alneberg]

I've actually haven't been using it. But I guess @chrisquince is better to answer if they are using it? Our experience was that it only marginally (if at all) improved the clustering for real data. But that could be due to the data we typically have as well.