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 think today is that day that I genuinely found something new that is in agreement with, and extends, previous research. Today is a good day!

Reading through old letters from Crick to Watson about the manuscript that would become "The Double Helix" [0]. The well worded, thorough, and scathing rebukement of Watson's book is impressive.

"If you publish your book now, in the teeth of my opposition, history will condemn.you, for the reasons set out in this letter."

[0]: https://profiles.nlm.nih.gov/ps/retrieve/ResourceMetadata/SCBBKN

Always like finding new command line utilities to use as a part of my workflows.

https://kadekillary.work/post/cli-4-ds/

"join" is a new one for me that I will definitely be using

I've recent stumbled upon Ben Langmead's teaching materials for lessons on bioinformatics, genome sequencing, and sequence alignment.

http://www.langmead-lab.org/teaching-materials/

There's a lot of really thorough material in there, while still remaining accessible to non-experts. Would highly recommend these lessons to anyone in/interested in the field

As of today, with Visual Studio Code Insiders, you can use your local installation to work on files on a remote server through an SSH connection. This is going to make it so much easier to build and analyze data on a computing cluster

https://code.visualstudio.com/docs/remote/ssh

Bowtie2 added official support for unaligned BAM files as input. I suspect that this will help start the move away from FASTQs as a data standard format and more towards the BAM format

https://github.com/BenLangmead/bowtie2/releases/tag/v2.3.5.1

For anyone who's interested in creating their own #conda packages, I've put together a brief intro on how to make them [1].

Hopefully it's a bit easier to follow than the official documentation for getting started. Any feedback is gladly welcomed

[1]: https://jrhawley.github.io/2019/03/28/conda-packages-intro

https://doi.org/10.1016/j.cels.2019.03.001

Couldn't. Agree. More.

It's unfortunate when the take home message from a talk is that you just need to read the paper, instead, because you didn't understand anything the speaker said. Kind of defeats the purpose of attending the talk

Keeping your conference talk on time shows respect for the other speakers and attendants. It's one thing to be slightly over time, but it's another to have >5 minutes of content over your allotted time

Technology company talks to a select group of PIs about a new product feels a lot like a committee meeting. Handling difficult questions, wanting everyone to like your work, and trying to justify your value

Precision recall curves don't define a single machine learning classifier, they define a class of classifiers parameterized by some threshold. It's important to not confuse a "good auPRC" value with a "good classifier", because there are lots of bad classifiers on that curve #KSepigen