Hi everyone
I'm not sure if this package is being actively maintained anymore, and so wasn't sure who to email, but I was wondering if you might be interested in integrating a tangermeme as a backend for some of your functions. https://github.com/jmschrei/tangermeme/
tl;dr is that tangermeme does all of the sequence manipulation / model handling that isn't the model itself, e.g., dinucleotide shuffles, inserting motifs into sequences, running marginalization experiments, calculating attributions, etc. It explicitly does not involve model training or performance evals like you have implemented here so I think they are complementary. It could significantly expand the scope of selene and/or reduce the maintenance burden by replacing some of your built-in functions.
Let me know if you'd like to chat more!
Hi everyone
I'm not sure if this package is being actively maintained anymore, and so wasn't sure who to email, but I was wondering if you might be interested in integrating a tangermeme as a backend for some of your functions. https://github.com/jmschrei/tangermeme/
tl;dr is that tangermeme does all of the sequence manipulation / model handling that isn't the model itself, e.g., dinucleotide shuffles, inserting motifs into sequences, running marginalization experiments, calculating attributions, etc. It explicitly does not involve model training or performance evals like you have implemented here so I think they are complementary. It could significantly expand the scope of selene and/or reduce the maintenance burden by replacing some of your built-in functions.
Let me know if you'd like to chat more!