Hacker News new | comments | show | ask | jobs | submit login
Metabolic Pathways (gatech.edu)
108 points by Tomte 11 months ago | hide | past | web | favorite | 24 comments

If you enjoy browsing pathways...


Thanks. Really helpful, if I forgot something and need to just refresh about the topic.

Now that's a complex one!

I found this paper a while ago and have it on my list of things to model: http://arep.med.harvard.edu/pdf/Chen99.pdf.

Ultimately, we could have a full gene-expression-as-a-dynamical-system model which we could perturb and use to study the dynamical features of genetic networks.

Anyone interested in this stuff?

This is a whole field full of multiple PhDs and ongoing research. It's called systems biology.

You should check out the OpenCOBRA project: https://opencobra.github.io/cobratoolbox/latest/

Is this substantially different from the one published by Roche?

It looks like a scan of the Roche "Biochemical Pathways" poster, but with the Roche logo and some text removed from the upper left corner. I used to have the physical poster; it's a very different experience studying the giant poster versus zooming in with a browser.

And we think we're going to hack this thing?

I think visualizing Linux kernel development can give a little perspective: https://www.youtube.com/watch?v=5iFnzr73XXk

If you view it in a single graph... yes, it seems completely intractable. If you can break it down into subcomponents and track the dependencies, it becomes much more tractable.

The thing with biochemistry is that it is highly connected, and very complex. Molecules aren't just products, they can also bind to gene regulators, or stabilize protein complexes. You can't cleanly separate parts of the system.

Reminds me of a talk by Andrew Endy[1], a long time ago, about synthetic biology. The essence is that his vision is to create "building blocks" of bio that have standard interfaces so you can then "code" them instead of untangling this spaghetti of life.

[1] - https://www.youtube.com/watch?v=xJFqqxxtbRY

This only means we need better tooling and better abstractions to be able to tackle this problem.

Difference with the linux kernel one is that this one is rather uncomplete. There are a lot of biological pathways that are just unknown.

To be fair, it looks like it could use a bit of refactoring.

More like a full rewrite. There is too much legacy, it is too difficult to read and the original developer is not responding to queries.

Typical case. The original developer has abandonded the project, maintainers make a complete mess of it, project ends up electing Trump as president of the United States.

I wish there was a jenkins variant that would reject because of architecture violations.

PullReuest was Rejected: No Singletons in this Component. No Singletons in this project.

What could go wrong? /s

It's too bad there's no virtual environment for the homo sapien OS.

There have been several attempts to virtualize cells, and come up with expressive PLs for hacking metabolic pathways. Turns out shit is too complicated

Aw, hell yes.

If you write very nicely to Roche they'll likely send you a copy which looks very impressive if you've room to pin it up.


The image has "unaligned seams" multiple places here. Looks like it was patched together from multiple images.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact