software engineer @arcadiascience, runner, cyclist, conifer aficionado.

Joined March 2009
12 Photos and videos
2 Dec 2023
Pytorch meets baby worms! It was a pleasure to work with @dave_matus and everyone at @ArcadiaScience to help bring this across the finish line (and on a tight deadline)!
Want to level up your automated phenotyping with DL/ML? At @ArcadiaScience, we’ve built a classifier to categorize developmental stages of #Celegans from high-throughput bright-field time-course images #WormClassifier 🧵1/12 bit.ly/nematode-classifier
5
11
3,191
Keith retweeted
Interested in computationally exploring how proteins are related across species? Join us @ArcadiaScience and engage collaboratively to design, create, and iterate on workflows and tools for comparing proteins in diverse organisms. Check out the job ad ➡️jobs.lever.co/arcadiascience…

2
16
28
25,085
Keith retweeted
Just launched v0.2.0 of our cell tracking tool 🚀 Check out our GitHub repo: github.com/royerlab/ultrack! Seamlessly combines segmentation & tracking via Python API 🐍 Scales effortlessly to handle massive datasets, like the #zebrahub paper. #CellTracking #Bioinformatics
6
79
314
42,520
Keith retweeted
📢 Very excited to see our work #cytoself now published in @naturemethods . #cytoself is a deep learning method for fully self-supervised protein localization profiling and clustering. Led by @liilii_tweet with @kchev @LeonettiManuel at @czbiohub nature.com/articles/s41592-0… 1/n
6
64
229
Keith retweeted
Check out our latest paper on self-supervised ML!!! w/ @loicaroyer @liilii_tweet @kchev, built upon OpenCell data. Loic’s tweetorial has all the info 👇
📢 Very excited to see our work #cytoself now published in @naturemethods . #cytoself is a deep learning method for fully self-supervised protein localization profiling and clustering. Led by @liilii_tweet with @kchev @LeonettiManuel at @czbiohub nature.com/articles/s41592-0… 1/n
2
8
29
23 Mar 2022
This should be required reading for new grad students (not to mention their PIs)
If you go for broke you often end up broken. If you swing for home runs you often end up striking out. But if you just put the ball in play—over and over again—good things tend to happen. On how to be consistent in work, sport, creativity, and really, all of life:
23 Mar 2022
Love the hierarchical colormap! 👌🏻
Replying to @SikkemaLisa
To build the core of the HLCA, we collected 14 published and unpublished datasets of healthy lung nose from 107 individuals, with cell type annotations, and technical and demographic metadata (age, BMI, etc.). We harmonized all metadata for downstream use. (4/20)
1
22 Mar 2022
This is a fascinating tour through the iconography of our scientific cultural heritage! 😍
Some scientific illustrations become memes*, some turn into cultural icons. It requires a combination of originality, aesthetic, simplicity and scientific accuracy. Here are some of them 🧵 1/n *memes in the original sense: units of cultural inheritance that replicate and evolve
1
21 Mar 2022
Perhaps a more specific name like `!ugly-hack-added-for-specific-regulatory-reason` would have helped to prevent its widespread (mis)use for other purposes 🤔
CSS co-designer here. !important was added for one reason only: laws in the US that require certain text to be in a given font-size. !important stops the cascade from changing it. Anything else is probably misuse, and a sign you may not understand the cascade properly.
1
15 Mar 2022
I'm sure BA.2 is coming but this kind of data seems to be *very* noisy esp when prevalence is low, and given the abundance of blue dots mixed w red dots, I would suspect this is mostly noise
15 Mar 2022
Wastewater surveillance of Covid is relatively sparse in the US, but 62 (15%) sites have a >1000% increase in viral RNA detected in the past 15-day period covid.cdc.gov/covid-data-tra…
1
11 Mar 2022
It has been a privilege and pleasure to work with so many amazing scientists @czbiohub to help reach this milestone! All of our data is freely available to explore and download at opencell.czbiohub.org - check it out and search for your favorite protein!

Our paper is out today in @ScienceMagazine, I’m stoked @OpenCellCZB ! Check out opencell.czbiohub.org: CRISPR tagging 3D live imaging IP mass spec to map the human proteome. science.org/doi/10.1126/scie… 👇 tweetorial from our 2021 pre-print.
2
14
Keith retweeted
#OpenCell is mapping our cells one protein at a time and paving a new way to visualize cellular architecture through a free, open-source website. Explore it now: opencell.czbiohub.org/ youtube.com/watch?v=6j4fpV-q…

1
10
20
Keith retweeted
Published today in Science, a team from CZ Biohub and @MPI_Biochem combined genome engineering, confocal live-cell imaging, mass spectrometry, and data science to systematically map the localization and interactions of human proteins. 🧵 science.org/doi/10.1126/scie…
2
34
116
7 Jan 2022
charts from hell 😵‍💫
With Omicron, there "are clear indications this wave will differ substantially from previous ones,” writes @ShamanJeffrey, an infectious disease modeler and epidemiologist. nyti.ms/3fdQZAF
6 Dec 2021
Does anyone understand what’s being plotted here? Why does 2-dose drop to zero then go up?
Keith retweeted
COVID Update: One too many smart people has told me or said on TV this week that the pandemic is over. I offer this thread as a single, uniform response to everyone. 1/
551
3,279
9,854
1 Nov 2021
This is a great example of effective data visualization: clear, honest, and contextualized
Excess mortality update, through September. (Code: osf.io/5n964/)
2
Keith retweeted
I am seeking an Associate R&D Engineer for our amazing BioE team at @czbiohub. We work on a wide range of important problems related to human health spanning from basic cell biology, infectious disease research, low-cost diagnostics, and outreach in underserved communities.
1
17
15
5 Oct 2021
you know it's a bad day when the lead story in the NYT ends with this correction
3
Keith retweeted
📢 Job alert (please RT): my lab is looking for a PhD-level scientist to join us @czbiohub. We study virus-host interactions across different viral families using a combination of CRISPR/base editing screens, molecular virology and cell bio methods. apply.workable.com/j/97E9644…

1
32
41