Ming Tommy Tang

Chatomics! — The Bioinformatics Newsletter Why Subscribe?✅ Curated by Tommy Tang, a Director of Bioinformatics with 100K+ followers across LinkedIn, X, and YouTube✅ No fluff—just deep insights and working code examples✅ Trusted by grad students, postdocs, and biotech professionals✅ 100% free

I hope you've found this post helpful.

Follow me for more.

Subscribe to my FREE newsletter chatomics to learn bioinformatics divingintogeneticsandgenomics.ck.page/profile

5. 42basepairs 42basepairs.com/

4. Intervene: a tool for intersection and visualization of multiple gene or genomic region sets

bitbucket.org/CBGR/interv...

GitHub - gosling-lang/gos: A declarative interactive genomics visualization library for Python. A declarative interactive genomics visualization library for Python. - gosling-lang/gos

3. The goal of **g(r)osling** is to help you build interactive genomics visualizations with [Gosling](github.com/gosling-lan...). This package uses [reticulate](rstudio.github.io/reticulate/) to provide an interface to the [Gos](github.com/gosling-lan...) Python package.

GitHub - PhanstielLab/plotgardener Contribute to PhanstielLab/plotgardener development by creating an account on GitHub.

2. plotgardener is a genomic data visualization package for R. Using `grid` graphics, `plotgardener` empowers users to programmatically and flexibly generate multi-panel figures

github.com/PhanstielLa...

GitHub - crazyhottommy/scATACutils: R/Bioconductor package for working with 10x scATACseq data R/Bioconductor package for working with 10x scATACseq data - crazyhottommy/scATACutils

I used that to plot single-cell ATACseq tracks github.com/crazyhottom..., more examples rpubs.com/crazyhottom...

5 tools to visualize genomic datasets 🧵
1. Karyoploter bernatgel.github.io/karyoploter...

A pan-cancer compendium of 1,294 plasma cell-free DNA methylomes and fragmentomes enabling multicancer detection www.nature.com/articles/s4...

Chatomics! — The Bioinformatics Newsletter Why Subscribe?✅ Curated by Tommy Tang, a Director of Bioinformatics with 100K+ followers across LinkedIn, X, and YouTube✅ No fluff—just deep insights and working code examples✅ Trusted by grad students, postdocs, and biotech professionals✅ 100% free

I hope you've found this post helpful.

Follow me for more.

Subscribe to my FREE newsletter chatomics to learn bioinformatics divingintogeneticsandgenomics.ck.page/profile

Whimsical — Where great ideas take shape Whimsical is a collaborative whiteboard app for thinking and planning. Create flowcharts, mind maps, wireframes and more.

10. whimsical.com/

How to Draw Flowcharts With LaTeX | Baeldung on Computer Science Learn how to draw flowcharts with LaTeX/TikZ.

9. Latex www.baeldung.com/cs/latex-fl... if you are geeky

8. Lucidchart www.lucidchart.com/pages/ very similar to draw.io

7. DiagrammeR if you want to play with R rich-iannone.github.io/DiagrammeR/

GitHub - fredhutchio/inkscape-tutorial: A tutorial for the free Inkscape cross-platform vector graphics editor. A tutorial for the free Inkscape cross-platform vector graphics editor. - fredhutchio/inkscape-tutorial

6. Inkscape github.com/fredhutchio... I use it to make figures (it is free)

5. draw.io

4. Miro miro.com/aq/ps/flowc...

3. Adobe illustrator

Flowchart Maker - Create Online Flowcharts for Free | Figma Build diagrams quickly and easily with our Online Flowchart Maker. Share ideas, diagrams and processes with your team in realtime, in Figma, for free.

2. Figma www.figma.com/templates/f... recently bought by Adobe

Free Flowchart Maker - How To Create Flowcharts Online | Canva Create a flowchart with Canva’s free online flowchart maker. Learn what is a flowchart, flowchart symbols, customizable flowchart examples.

1. Canva www.canva.com/graphs/flow... it can also make a nice logo. I made my Linkedin banner with it.

10 tools to make a better flowchart than in PPT 🧵

Chatomics! — The Bioinformatics Newsletter Why Subscribe?✅ Curated by Tommy Tang, a Director of Bioinformatics with 100K+ followers across LinkedIn, X, and YouTube✅ No fluff—just deep insights and working code examples✅ Trusted by grad students, postdocs, and biotech professionals✅ 100% free

I hope you've found this post helpful.

Follow me for more.

Subscribe to my FREE newsletter chatomics to learn bioinformatics divingintogeneticsandgenomics.ck.page/profile

- Validate on truly independent data, not just held-out data from the same source
- If your model works "too well," be suspicious

The best ML in biology isn't just accurate.

It's accurate for the RIGHT reasons.

How to protect yourself:

- Always check what your model actually learned (use SHAP, LIME, GradCAM)
- Watch for confounding: are your labels correlated with batch, site, or technical variables?

It works amazingly well -- until it encounters conformations far from training data.

4/ AlphaFold and structure memorization.

AlphaFold is incredible. But research has shown that for fold-switching proteins, some of its predictions are driven by memorization of training-set structures rather than learned protein physics.

Substantial batch effects in TCGA exome sequences undermine pan-cancer analysis of germline variants In recent years, research on cancer predisposition germline variants has emerged as a prominent field. The identity of somatic mutations is based on a reliable mapping of the patient germline variants. In addition, the statistics of germline ...

3/ another example: Substantial batch effects in TCGA exome sequences undermine pan-cancer analysis of germline variants

pmc.ncbi.nlm.nih.gov/articles/PM...

It Shows that deep learning models trained on TCGA histology data can infer submitting site rather than actual biology, inflating prognostic accuracy.
www.cell.com/cancer-cell...

2/ cancer predictions

Howard et al. (2023) - "Multimodal deep learning: An improvement in prognostication or a reflection of batch effect?" Cancer Cell.

But many were learning hospital-specific artifacts -- image contrast, positioning (AP vs PA), and even hospital labels embedded in the images.

The model learned WHICH HOSPITAL the X-ray came from, not what COVID lungs look like.

1/ COVID-19 chest X-ray classifiers.

Multiple deep learning models claimed to detect COVID from chest X-rays with high accuracy.