The objective of the proposed research is to devise advanced detection algorithms for the read channel in magnetic recording. To cater for the explosive demand for data storage, the magnetic recording industry keeps pursuing a higher areal density, packing more bits per unit area on the disk. The resulting increase in media noise and other interference requires the detection techniques to keep pace. To combat multiple sources of interference, a state-of-the-art iterative detector in magnetic recording has numerous parameters including equalizer coefficients and pattern dependent parameters like signal levels, predictors, and residue noise standard deviation. The number of parameters will further increase with the push towards higher areal densities. There is thus a need for an efficient and effective tuning strategy. We propose three schemes: (1) an adaptive BCJR tuning method to minimize the frame-error rate; (2) a time-varying detector to maximize the channel capacity; (3) a frame-dependent detector that adapts to the detected frame during turbo detection. The performance of an iterative read channel is dominated by two choices: how to choose the detector parameters, and how to choose the error-control code. While prior work has optimized two separately, no one has put them in the same picture. Therefore, we propose a strategy for simultaneously optimizing the detector and the error-control codes. The approach is based on the proposed time-varying detector that maximizes the channel capacity. An optimized LDPC code could then be designed correspondingly. Furthermore, we for the first time try to introduce transforms into the read channel. Two promising transforms, Hadamard and Haar, are expected to help mitigate the media noise. The design of the transform-based detector is in progress.