What are your go-to resources for scientific visualization? Do you have a favorite chart type that everyone else seems to hate? (I’m looking at you, pie charts... 🥧)

#Bioinformatics #OpenScience #DataViz #ScientificCommunication [5/5]

Even after years of making figures, I’m still a student. 🎨

There is a specific thrill in making a figure that is both mathematically "informative" and aesthetically "beautiful." Clarity is a competitive advantage in science. [4/5]

From data to Viz | Find the graphic you need A classification of chart types based on their input data format.

For daily implementation, I rely on "From Data to Viz." 🛠️

It’s a brilliant resource that provides:
✅ A navigable flowchart to pick the right plot.
✅ Code snippets for quick execution.
✅ Warnings about common "caveats" for each chart type.

Check it out: data-to-viz.com [3/5]

My north star for visualization principles is the "Points of View" series in Nature Methods by Bang Wong and Martin Krzywinski. 🧪

They bridge the gap between design theory and complex biological data. If you haven't read them, start there. [2/5]

The lifelong craft of scientific visualization

In data-driven R&D, your model only matters if your stakeholders can understand it. 📉
Communicating insights is perhaps the only part of the process that truly "counts" at the decision-making level.

But great #DataViz is harder than it looks.🧵 [1/5]

Is it "Data Science" if the most critical step—the generative model—relies on intuition and domain feel? Or are we just high-tech artists painting with numbers? 🖌️

What do you think? Is the "Art" of modeling being overlooked?

#DataScience #Bioinformatics #Bayesian #Modeling #CompBio [5/5]

This requires "Domain Fidelity"—using intuition and deep expertise to pin down what the data should look like before we ever run an analysis.

Counter-intuitively, this is almost never taught in standard curricula. It’s learned on the job, through trial and error. [4/5]

The "Art" part is building the Data-Generating Model. 🏗️

How do we ensure our synthetic or simulated data actually resembles the messy, high-dimensional reality of a biological system? This is much more amorphous. [3/5]

Effective modeling requires two entities:

The Data 📊

The Compendium of Algorithms ⚙️

The "Science" part is matching the algorithm to the data. We’re taught this in every stats and engineering course. It’s systematic. It’s logical. [2/5]

Data Science or Data Art?

Building predictive models for biology and medicine is often sold as a "pure" science. 🧬

But in practice? It’s as much about art as it is about algorithms.

Here is why I think we might need to rename the field to "Data Art." 🎨 [1/5]

The bottom line: Benchmarking—much like success in life—is not a zero-sum game.

It’s about finding the right tool for the right job, not finding the "one tool to rule them all." ⚔️

What’s your "litmus test" for a trustworthy benchmark?

#Bioinformatics #OpenScience #DataScience #CompBio

To maintain integrity, I’m a proponent of treating benchmarks like clinical trials. 🛡️

Document decisions, metrics, and datasets upfront—resembling a pre-registration (as suggested by Brooks et al., 2024). This keeps the analysis actionable and honest.

We need to stop looking for a single optimum and start seeking the Pareto Front. 📈

Benchmarking should be a nuanced elaboration of where and when benefits are had—and, crucially, where they are NOT.

Real-world performance is rarely a straight line.

The fear: If a new method is worse in even one metric, it’s seen as "insufficiently novel."

The result: Cherry-picking metrics and datasets to show a method in its best light. It’s effectively p-value hacking for benchmarks.

Users suffer because they lose clear guidance. 📉

Reading the vast majority of computational biology studies, you’d think every new method is categorically better than the last.

Statistically? A highly unlikely scenario.

We’ve fallen into the "Novelty Trap," where we view benchmarking as a zero-sum game. ⚖️

Very interesting! Congraluations !
I saw an example of Fitz-Nagumo on github -- does this in general work also for large nonlinear systems?
Do you plans to have implementations in other languages (😉 Python?)

This is brilliant -- great work. Cant wait to try this out on my favorite feedback systems. Congratulations.

🦋✨ #FluorescenceFriday meets developmental biology!
This glowing "butterfly" isn't an insect, it's a transverse section of a developing chicken gonad. When art meets science, even sex differentiation takes flight! Image taken by @martinestermann.bsky.social #ChickenModel #DevBio 🐥🔬

ML uncovers 24h rhythms in cancer 🕒🧬
Using COFE ☕, we reconstruct #circadian rhythms from human tumor biopsies across 11 #cancer types!
Turns out, tumor #clocks tick in a different timezone 🕰️💊
Got data? Let’s connect! 🤝
#AIforHealth #CircadianMedicine #oncology

🚀 Thrilled to collaborate on this study led by Jihwan Myung, where we explored why organisms prefer #networks of autonomous #clocks over a single central #oscillator, proposing enhanced robustness, specialization, and adaptability.

👉 buff.ly/xhMsboJ

🙏 @dfg.de
#circadian #circadianmedicine

Enhance your experience of reading scholarly articles by taking advantage of Google Scholar's PDF viewer. Set it as your default PDF reader in Chrome to seamlessly access AI-created summaries for each section. Try it out here: buff.ly/LjX2K6d #research #tips

Do you know a better tool? Lemme know.

StuROPx bringt Studentinnen aus ganz Berlin zusammen ❤️! Danke an das #BUA und @humboldtuni.bsky.social für die Möglichkeit und Unterstützung!

Circadian PERIOD proteins sculpt the mammalian alternative splicing landscape Mammalian circadian oscillators are driven by a transcription-translation feedback loop where CLOCK:BMAL1 activity is repressed by the PER:CRY complex. While transcriptional regulation by PER is well…

🚨New preprint alert!🚨 Our latest study (with K. Padmanabhan & Y. Suzuki) on post-transcriptional #circadian regulation is out.

🔬 Key finding: #Clock proteins regulate #splicing of non-rhythmic transcripts too

Read it here 👉 buff.ly/xsanmqw

#omics #ScienceTwitter #Nanopore #multiomic

Celebrate Pi Day and Holi (Festival of colors) by diving into the vibrant world of Penrose tiling! Combine the joy of coloring with your little ones with the beauty of #math and patterns (buff.ly/8RR568J).
It's a colorful and fun way to explore and learn! 🌈📐 #PiDay #HoliFestival #idm314 #STEM

It's #BrainWeek! My journey in #circadian research began with the brain's internal #clock, initially thought to solely manage 24-hour body cycles. Now, we see it as a "first among equals," as most organs have their own #clocks. With regular rest & ample sunlight, the clock can be your #health ally.

This book argues that generalists excel over specialists in solving wicked real-world issues. Diverse perspectives and creativity are key when there’s no clear optimal solution. It’s changed my view as a teacher, researcher, and parent. Highly recommend! #Generalist #ProblemSolving #CreativeThinking