Add Cytochrome P450 inhibition ML model

[ikmckenz]

SwissADME includes a Cytochrome P450 inhibition model in their output, their model has outputs for five specific enzymes. Their model is an SVM that gets 70-80% accuracy, so nothing too crazy. We could implement a similar or better machine learning model.

[soulclimberchick]

Hey there @ikmckenz. I happen to be working on my capstone for my data science/machine learning Bootcamp. My capstone idea was to build a model around Cytochrome P450 inhibition and I somehow serendipitously found myself here when looking up how to get the right type of data out of Chembl on Reddit. I would LOVE to contribute this exact thing as part of my capstone. I may need a little guidance along the way for things here and there just because I'm diving into the bioinformatics side head-first.

I happen to know a lot about the CYP450 block of enzymes and have a vested interest in furthering research. Would you be willing to give me pointers here and there specific to working with bioinformatics data if I run into snags that my instructors can't help me with so that I may contribute?

[edit] I should note, that I have contributed to open source projects on GitHub in the past so I'm not a total n00b.

[ikmckenz]

Hey @soulclimberchick. What an excellent coincidence!

I would be very happy to guide you along with anything you may need.

[soulclimberchick]

Awesome, thanks for getting back to me! I finally found the CYP450 dataset that was talked about in the SwissADME paper here: http://www-metrabase.ch.cam.ac.uk/download.html. I downloaded the newer version which was not used in the original models. They have a breakdown of the two datasets, I'm curious to see if the larger dataset will be better or just add more noise. We shall see!

Do you have a preferred way to collaborate back and forth? We can work within this ticket or tickets which may not be the best place for chatting back and forth, or we can do email or something else. Let me know and we'll make it happen!

Kira

[soulclimberchick]

@ikmckenz I will be wrapping up this project by the 21st of this month at the latest (my presentation is on the February 22nd or 23rd and my final code book is due by 5pm PST on the 23rd. I'd love to connect with you as soon as you are able so that I can make sure that I can get a little help and making sure that what I am thinking of doing makes sense for this specific piece of the puzzle. Let me know how best to reach you and when you have some availability!

Kira

[ikmckenz]

@soulclimberchick found you on linkedin and messaged you there.

[DarthSaddest]

Any progress on this?

[ikmckenz]

Code from @soulclimberchick never came through, we're still open to pull requests!

[soulclimberchick]

My code should all be pushed in my branch, it was just never merged. I haven't taken the time to refactor things for production, though.

…

On Apr 22, 2022, at 11:04 AM, Ian McKenzie ***@***.***> wrote:  Code from @soulclimberchick never came through, we're still open to pull requests! — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.

[soulclimberchick]

Hey @ikmckenz I am in between jobs and would love to work on productionizing this with you when you have some time. I'm so sorry I kinda disappeared. I ended up over-committing myself for a long time and then I was just in a state of general overwhelm until prettymuch this week heh. It was rough. But I think about this project often, and really would like to help clean it up and keep working on it. Even if it's little by little. I'll try to set aside some dedicated time each week to chip away at it without overcommitting myself again. :)

[ikmckenz]

No problem! Let's get your model productionized.