-
Notifications
You must be signed in to change notification settings - Fork 45
Comments
Conversation
I add Epitope data from tdc here : https://tdcommons.ai/single_pred_tasks/epitope/ I will need help in validation my approach, need to ensure the indices start with 0 or 1 in epitope active binding one
for more information, see https://pre-commit.ci
kjappelbaum
commented
Mar 11, 2023
@phalem Thanks a lot for your efforts; it is really great to have you take the lead on the datasets; we could only have that much progress thanks to your contributions.
One thing with your PRs is that they currently fail in our automatic checks. This is mostly the case because some lines need to be shorter. Could you break long (>120 character) lines to make the content more readable?
Do you think that makes sense? If you want help with that or to discuss this, I'm happy to do so.
phalem
commented
Mar 11, 2023
@phalem Thanks a lot for your efforts; it is really great to have you take the lead on the datasets; we could only have that much progress thanks to your contributions. One thing with your PRs is that they currently fail in our automatic checks. This is mostly the case because some lines need to be shorter. Could you break long (>120 character) lines to make the content more readable?
Thanks kevin for your kind words. I think that problem raise from description that TDC provide. This can be solved by break target description and URI links, other things follow the same workflow. Please, tell me if you have time now to do that during review or I should do by myself. If this doesn't work, we will need to see if description need to be summarized
for more information, see https://pre-commit.ci
@phalem
phalem
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Add benchmark field
data/iedb_jespersen_et_al/meta.yaml
Outdated
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
so the target is a sequence and the input is the full sequence?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
so the target is a sequence and the input is the full sequence?
Yeah, I forget to comment on that. Input was sequence and target is sequence as well we could make it categorical, but with a lot of category. I don't know what is your opinion of that.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, this is how it is currently implemented. I set it up to only give the corresponding epitope sequence without the _ for the other parts of the sequence.
@kjappelbaum
kjappelbaum
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Once more: Thanks 💯
(And similar comments)
As data was complex and get in an indirect way. I didn't implement split
phalem
commented
Mar 29, 2023
Before I forget @kjappelbaum we need someone to ensure if index of start with 0 or 1. As it will be different sequences and situation
kjappelbaum
commented
Apr 27, 2023
@MicPie we need to carefully look into this. I don't think this is a simple classification/regression task.
kjappelbaum
commented
Apr 27, 2023
Also need to check how this overlaps with #67
I add Epitope data from tdc here :
https://tdcommons.ai/single_pred_tasks/epitope/
I will need help in validation my approach, need to ensure the indices start with 0 or 1 in epitope active binding one