#Genetics #Genomics #Cancer #Academia #AcademicChatter
Interested in cancer genetics? Don't want to fly somewhere to here about it? Next week the American Society of Human Genetics is having an online-only digital symposium on cancer genetics.
https://www.ashg.org/product/2025-spring-symposium-emerging-frontiers-in-cancer-genetics/
I think it's also worth up to 6 #CME credits if you do continuing medical education.
#Genomics
#Bioinformatics
#Academia
Beware US academic coders. If you have collaborators contributing code from a sanctioned region, your repos may be locked.
See https://mastodon.social/@organicmaps/114155428924741370
The kicker is that the US is leaning authoritarian and all in on censorship. If Canada or Mexico become sanction targets and you have code pushes from there, Microsoft will lock you out of your GitHub accounts.
In today's news, CEO of AI company grasping for broad relevance claims that AI has to be everywhere or you'll be left behind.
#Academia
#AcademicChatter
#Genetics
#Genomics
Work in genetics and genomics? Do you sequence things? Are you the go-to in your department or division when people don't know what to do with sequencing data?
We reworked our cold archive to zstandard instead of gzip. It's substantially smaller but takes more resources up front. You may want to think about switching archive strategies.
I rarely set aliases (just because it makes me forget full commands when I move to new systems). I did set this alias on the the cluster to make it easier to remember.
alias safe_rsync='rsync --exclude .Rproj.user/ --exclude .git/ --exclude .snakemake/ --progress --verbose --recursive --update'
Depending on the situation I'll throw in --size-only, but I prefer not to when possible.
I often rsync files between different locations. For example, between scratch space and active storage on the cluster, or between the cluster and local servers.
Each project is a Git repo with R project info and usually with Snakemake workflows, giving me .git, .snakemake, and .Rproj.user directories.
At new locations, I pull the git repo first. I don't want to copy those dot directory contents between locations. Which means that when I rsync I want to ignore those directories.
US citizen that doesn't love protests or big crowds? Like computers? Love data? Contribute CPU time to the Internet Archive backing up government data.
https://github.com/ArchiveTeam/usgovernment-grab/blob/master/README.md
Their presentation was about 85% based on a project I did more than half the analysis work on. 100 people ze PI spoke to about it along the way were thanked. We get to the last with all the contributing people spelled out.
Guess who was completely missing? I bet you can guess.
I get really frustrated with the US academic research ecosystem sometimes. Or often. Probably often. In particular how translational genomics and bioinformatics is perceived.
I've spent more than a decade helping every place and time I was asked to.
"Don't know analysis? Happy to help"
"New student? Happy to help."
But when the research leads to money, suddenly the budget can't support my help. This week was a great example. I listened to a talk this week from someone I helped in the past.
Friends. Remember to try and make your outputs self documenting. I was just asked for methods for an analysis from *4 years ago*. This was before I moved to singularity with docker images.
Random code: redirected output logs show versions.
R code: I always include session info at the end of every script. Which includes the R version and package versions.
All located within about half an hour.
Fabric posters are pricier, but totally worth it for me. Sent PDF in on Tuesday. Got the poster in the mail a minute ago. Never had any issues with Postersmith. Easy to stuff in my bag with no poster tubes. Easy to stow in my backpack on the day I have to put it up and later in the day to take it down.
#academicchatter
#academia
#ASHG24
Big business dudes. My guys. If your business model is "wholesale intellectual theft" and "enormous power demands" for something a search tool did better in 2005, it's a failure. Your model is a failure.
Instead of pitching that climate targets are unimportant because you want the power at any cost, what if instead we spent these enormous amounts of money on decarbonizing transportation and the electrical infrastructure?
I even have a plan to pay for it. It's called "corporate taxes."
Modern #AI pitches just kill me. There are areas where machine learning generally and artificial intelligence specifically can be useful, particularly in medicine. Helping screen images, extracting data that isn't in regular formats, etc.
The big business pitches are all-in on generative AI and #LLMs. They are by and large not that useful in their current forms. Now those same companies who create not very useful products want to run *NUCLEAR POWER PLANTS* to serve their sub-par products.
Ah yes. Cory Doctorow on the enshittification of Epic for health records.
https://mamot.fr/@pluralistic/113237780216406333
#EHR
#ElectronicHealthRecord
#MedicalSoftware
#Enshittification
Our health system bought Epic a few years ago. It has been hilarious and frustrating as a non-clinician to see how many conversations along the lines of "what do you mean it can't do that???"
Our network & information security people emailed in the last week to warn that using the Adobe Acrobat built in #AI assistant on PDFs that contain protected health information shares it with Adobe and is forbidden
It's almost like a stuffing data snaffling privacy nightmare feature into all software on everyone's computer in an environment with protected information is a bad idea. Who knew!
Recruiters are wild on LinkedIn.
Hey! We represent [company] and you look like a perfect fit for [job that I could definitely do].
OK. I look at the job ad. It lists salary (a plus!) that is... essentially not different than current salary, would require moving to a large city in California, would be for a company "too fast paced" to be 9-5, but you also get "unlimited vacation!"
Riiiiight. Move across the country to not make more on paper before really not making more in real terms. Nope.
This is your periodic reminder that public #libraries are amazing infrastructure.
A few weeks ago we were in another state with no Internet access or cell service where we were staying. During that trip my Dean wanted me to do a large group Zoom conference.
The local public library let me reserve a room with Wi-Fi access for an hour. Super easy. Set it up online. Took 5 minutes to set myself up.
3) trusting the outputs. It's too close to opinion for bioRxiv and medRxiv. Why tell me you wanted the piece if you didn't? I literally went through and used the outline I sent you. Jiminy crickets. Total waste of my time and clearly that audience will just truck along making the same poorly judged decisions in their clinics. So frustrating.
2) It was suggested I write up the dangers for Arthritis and Rheumatology. I query the editor. They want a list outline. I send it. They want me to write it to send to them. I do. Days later:
We regret to inform you this isn't of interest. It only briefly covers uses in rheumatology. Not of interest.
Ok... You missed the point. The point is that how you WANT to use them and how they work are at odds. You don't understand the risks of stuffing patient data into a corporate model or blindly