You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I was reading your paper with excitement earlier today but was disappointed by the widespread underperformance in perturbation outcome prediction. Our Dictys method first infers a direct GRN consisted of the direct edges that require TF binding. But to predict perturbation outcomes, a total-effect GRN needs to be computed through infinite propagation of the direct GRN. This is described in our Fig 4e-g and implemented in https://github.com/pinellolab/dictys/blob/cf779bb441acd41aadba7b13aeba8abae2f671e7/src/dictys/scripts/makefiles/common.mk#L161-L165 . I did not find similar lines in your repo.
Therefore I'm curious whether you had any chance to test the total-effect GRNs from Dictys when comparing against knocktf? If not, that is totally fine because I already plan to use greta to improve Dictys, during which I can try it out. However, if you are also interested and happen to find that Dictys helps, it would definitely bring a more encouraging picture to the field! I am happy to help with troubleshooting if needed.
The text was updated successfully, but these errors were encountered:
We are preparing to submit a new version of the preprint today or tomorrow (with more methods and metrics) so I do not have time to include this right now, but for sure for the next iteration! 😉
BTW, feel free to do open a PR to add these lines, else when I'm back from holidays I'll do it.
I haven't dived into you code deep enough to figure out how to change the GRN only for knocktf but not for other evaluations. Maybe you prefer to keep things clean and avoid a dummy method dictys-total? Guess it's best to leave that question to you and not create a PR myself.
Anyway, thanks for the quick reply! Best of luck with the submission and enjoy your well-deserved holidays!
Congrats @PauBadiaM on your PhD and GRETA!
I was reading your paper with excitement earlier today but was disappointed by the widespread underperformance in perturbation outcome prediction. Our Dictys method first infers a direct GRN consisted of the direct edges that require TF binding. But to predict perturbation outcomes, a total-effect GRN needs to be computed through infinite propagation of the direct GRN. This is described in our Fig 4e-g and implemented in https://github.com/pinellolab/dictys/blob/cf779bb441acd41aadba7b13aeba8abae2f671e7/src/dictys/scripts/makefiles/common.mk#L161-L165 . I did not find similar lines in your repo.
Therefore I'm curious whether you had any chance to test the total-effect GRNs from Dictys when comparing against knocktf? If not, that is totally fine because I already plan to use greta to improve Dictys, during which I can try it out. However, if you are also interested and happen to find that Dictys helps, it would definitely bring a more encouraging picture to the field! I am happy to help with troubleshooting if needed.
The text was updated successfully, but these errors were encountered: