Compressed beam-selection (CBS) exploits the limited scattering of the millimeter wave (mmWave) channel using compressed sensing (CS) and finds the best beam-pair with limited overhead. The CBS procedure can further benefit from the knowledge of some additional structure in the channel. As mmWave systems are envisioned to be deployed in conjunction with sub-6 GHz systems, we use the spatial information extracted at sub 6 GHz as out-of-band side information about the mmWave channel. In particular, we formulate beam-selection as a weighted sparse signal recovery problem, and obtain the weights using sub-6 GHz angular information. Furthermore, we formalize the notion of spatial congruency between sub-6 GHz and mmWave, and numerically evaluate the degree of spatial congruency necessary for the success of the proposed approach. The simulation results illustrate that the proposed approach reduces the training overhead by 3x compared to the CBS approach.