James Hawley @jrhawley@scholar.social

I'm a PhD candidate in Medical Biophysics at the University of Toronto studying epigenetics and cancer. Math and physics background taking a step into bioinformatics and evolution #introductions

I've just discovered that Snakemake does HTML reports as of v5.1...this is amazing

https://snakemake.readthedocs.io/en/stable/snakefiles/reporting.html#

There's even a brand new Bayesian algorithm the authors developed to more accurately detect early and late RNA, which they also introduce with a single sentence.

Those two things alone must have taken an unbelievable amount of time and they don't even take up 2 sentences in the entire paper

To be clear, I don't blame the authors for this. It just amazes me that Nature does this and forces scientists to write such extremely succinct letters that obviously contain way more information than can reasonably be described in 5 pages

The authors are much more extensive in the supplementary information (20 pages long), but to shorten an entire method development process to not even half a sentence seems almost criminal

This has to be the most glossed over description of a methodological breakthrough I've ever seen.

https://www.nature.com/articles/s41586-019-1369-y

The entire paper is possible because of adapting a bulk sequencing experiment for single cell sequencing. This probably took months if not years of work, and the authors describe that process like this:

> After optimization for single-cell sequencing...

Kudos to the Anaconda team for conda v4.7. Solving dependencies and installing packages is now much _much_ faster than it was in v4.6

Here's a really good explanation of where PCR duplicates come from in the sequencing process.

http://www.cureffi.org/2012/12/11/how-pcr-duplicates-arise-in-next-generation-sequencing/

With probabilities, graphs, and a good explanation at the end of why PCR is even necessary

Nearly every single genomics paper requires sequence alignment, and I can't count the number of times I've seen Bowtie or BWA mentioned in the text and no citation is provided.

These are some of the best software tools in bioinformatics available today, and people use them because of their thoroughness, flexibility, and ease of use.

Those things takes years and years to perfect, and that work deserves recognition

You need to reference these things in your papers.

Not only for reproducibility of your work (version numbers are very important), but to give credit to those authors for their hard work

Imagine if you spent years working out tedious details of an important problem in the field, implementing them in an easy-to-use and well-documented way, and someone uses your exact method and doesn't reference your work.

In any other context, that would be some form of academic misconduct

I had a collaborator ask me to write up my analysis for the methods section of a paper, and I explicitly described each step, including software tools, version numbers, and citations for each one.

I found out that the version he submitted to the journal kept the text the same but removed all citations to the tools, despite them being critical to doing the analyses

This is probably more true than I'd like it to be

https://twitter.com/PhilippBayer/status/1137927141981544454

"So why build a research quality data set? ... this work will always be done, one way or the other. If you don’t invest in making a research quality data set up front, you will do it as a thousand papercuts over time."

https://simplystatistics.org/2019/05/29/research-quality-data-and-research-quality-databases/

Some good examples of why making Dockerfiles isn't always straightforward

https://pythonspeed.com/articles/dockerizing-python-is-hard/