[insilico]

Hello!

I am writing the specs to predict the properties of herbs, in silico.

This is the flow I worked out so far:
- Take one herb
- Run a bio-assay [How? Is there any standardized assay for herbs? Or one for each potential element?]
- Feed the chemical structure into a DB
- Check the docking [Not in absolute terms, but as likelihood %]
- Check bioavailability [Oral, dermal, etc.]
- See for what it is likely to be beneficial for [Check functional groups? Any other potential patterns?]

And these are some useful metaresources I found:
http://www.click2drug.org/directory_StructureBasedScreening.html
http://www.nature.com/nrd/journal/v12/n12/fig_tab/nrd4128_T1.html

Do you have:
- any feedback about the flow
- comments about the existing tools we can use to do this

THANK YOU SO MUCH!

[Arkcon]

This is a fairly good outline, and so is a good place to start on your project.  However, the devil is going to be in the details ...

Hello!

I am writing the specs to predict the properties of herbs, in silico.

OK, I have to ask right away: in silico -- I'm going to assume you're going to attempt to use cheminformatics to make predictions.  Its just that "in silico" is your username on this forum as well.  So I wanted to be clear.

This is the flow I worked out so far:
- Take one herb
- Run a bio-assay [How? Is there any standardized assay for herbs? Or one for each potential element?]

One herb?  OK, I can visualize that.  But standardized assays, for a herb, no that's not an apt term at all.  For each potential element -- OK, so you mean active pharmaceutical -- be that a molecule or class of molecules.  OK, there may be standards for some things, but not for others.  We'll need examples.  But I'm itchy already ... this is lready too general to work well, if at all.

- Feed the chemical structure into a DB

This part is clear.

- Check the docking [Not in absolute terms, but as likelihood %]

This bit of jargon, you will have to define better for us.  I have no idea what you mean.

- Check bioavailability [Oral, dermal, etc.]

Hmmm... why is this here?  Then again where else should it go?  I don't know, but I'd like you to justify your reason, just to make your thesis better.

- See for what it is likely to be beneficial for [Check functional groups? Any other potential patterns?]

Again, does this step go here, and not sooner?  Why?  And you want to determine the medical benefit, by functional groups?  That's not how pharmaceuticals are designed/discovered, so I think this is a way more complicated problem then you seem to think it is.

And these are some useful metaresources I found:
http://www.click2drug.org/directory_StructureBasedScreening.html
http://www.nature.com/nrd/journal/v12/n12/fig_tab/nrd4128_T1.html

Good, you have references, see how your plan matches theirs, and use that to justify your plan.  If it doesn't or you can't justify it, then consider adopting their plan at that step.

Do you have:
- any feedback about the flow
- comments about the existing tools we can use to do this

We'll have to see how other members weigh in.

THANK YOU SO MUCH!

A nice start, we'll have to see how it grows.

[curiouscat]

What kind of assay? Are you going to ignore the loads of prior info out there about what works where?

[insilico]

>>> curiouscat

Good point. However, the goal is to build a flow/platform which can be used for untested herbs as well. Other than that, yes, for herbs with well known benefits we can go straight to the DB part.

>>> Arkcon

Thank you for the detailed feedback, I really appreciate that!

>>> I'm going to assume you're going to attempt to use cheminformatics
Correct!


>>> One herb?  OK, I can visualize that.  But standardized assays, for a herb, no that's not an apt term at all.

Example: Rhodiola. I know there is already data about its chemical composition etc. But we also know that the type of soil etc. make a big difference. So, for a start, we want to know its active ingredients.

>>> Check the docking [Not in absolute terms, but as likelihood %] / Check bioavailability [Oral, dermal, etc.]

Clarification: we need to understand if the active ingredients can be digested or used in other ways by the human body. I think it makes sense to enter the data about a herb, even if not all the active ingredients are bioavailable. Maybe one day researchers find a way to make it available by adding specific enzymes etc.

>>> - See for what it is likely to be beneficial for //  That's not how pharmaceuticals are designed/discovered

Can you please provide 2-3 urls/specific kws that you consider very relevant?

The whole flow is very general, because it is meant to be applicable to virtually every herb/desired outcome. That's a bit of exponential vs. linear discovery. At this stage, I am not sure if exponentials will apply here, but considering the benefits for all if it does work, it is work investing part of my time in this.

I would not be the one taking over the massive workload by myself. I'll hire help if at least the flow develops into something applicable/valid. THANKS!

[curiouscat]

I think what's missing is validation. How will we know if your approach throws promising hits out or duds?

I'd say test on existing known molecules to see if the hits it identifies are indeed known to be good.