Reconfigurable intelligent surfaces (RIS) are one of the possible candidate technologies for 6th generation (6G) wireless communications owing to their robustness against weak channel conditions. They allow using of an additional reflecting surface to assist the information transmission between the base station (BS) and user equipments (UEs) to improve the communication system performance resulting in a more favorable communication environment. In this paper, an overall perspective for RIS-enabled channel estimation is presented where the channels are modeled as correlated (i.e., as the realistic case) due to the spatial deployment of transceiver antennas. Accordingly, two main channel estimation approaches are considered to determine the performance of the overall RIS-enabled wireless communication. These approaches include i) least squares-based conventional estimation for the effective channel consisting of a direct channel and RIS-assisted cascaded channel and ii) deep learning (DL)-aided estimation. Computer simulation results show that the channel estimation performance improves as the channel correlation coefficient increases and bit error rate performance enhances when the number of RIS elements increases. The presented framework is important in the overall evaluation of the channel estimation performance of RIS-enabled 6G communication systems.