eQTL_variant_effects_aracena_models

Context

Having trained several different models relating predicted Enformer variant effect predictions to ground truth Aracena signal, it is now worth testing these by observing their ability to predict the effect of genomic variants. Due to the way these models have been trained, the variant effects can be simulated by performing point mutations on enformer input sequence, running these sequences through enformer, and then passing the resulting predictions through my trained models. The resulting predictions can then be compared eQTL effects observed from the Aracena dataset. Should these predictions be accurate, it would suggest that the models have learned to predict the effect of genomic variants on gene expression much in the same way