Future wireless networks will be characterized by high network density and high communication frequencies. In such a scenario, this paper studies the resource allocation problem in Integrated Access and Backhaul networks operating at Terahertz (THz) that is a largely unexplored area. Resource sharing requires an efficient allocation strategy that is investigated here taking into account the re-irradiation phenomenon caused by molecular absorption characterizing THz communications. This effect induces spatial diversity that can be exploited through MIMO systems. The paper proposes a two-step allocation strategy aiming at maximizing the user sum rate: (i) first THz sub-bands are split between backhaul and aggregated access links, and then (ii) every IAB node allocates portions of its sub-bands to its associated UEs to satisfy their minimum rate targets. Step (i) is formulated as a combinatorial optimization problem and solved via a Genetic Algorithm, achieving near-optimal performance with significantly lower complexity than other solutions.