Voltage-gated potassium (Kv) channels are cell membrane proteins that allow a cell to register and transmit electrical signals throughout the nervous system. Malfunction in these channels can be linked to different diseases, such as epilepsy, sleep disorders and cardiac arrhythmia, to name a few. A subfamily of Kv channels, Kv7, has been identified as possible targets for future drug design. However, difficulties arise from similarity between subtypes within the Kv7 family of channels, which can have about 40-65% sequence identity and with key protein regions being 90% identical. Ideally, a potential drug should only modulate the activity of individual subtypes among the Kv7 family, but due to their close similarity it is difficult to find good candidates with traditional drug design techniques. Usually modulators tend to target all or most of the Kv7 subtypes and modulation of incorrect subtypes might results in adverse side-effects. While traditional design approaches have failed to produce potential drug candidates, deep generative modeling has in recent years emerged as a viable solution. Thus, the aim of this project is to build a generative model for sampling potential drug candidates that target individual subtypes among the Kv7 channels without affecting undesirable ones.