Some computational aspects and behavioral patterns of P systems are considered, emphasizing dynamical properties that turn useful in characterizing the behavior of biological and biochemical systems. A framework called state transition dynamics is outlined in which general dynamical concepts are formulated in completely discrete terms. A metabolic algorithm is defined which computes the evolution of P systems modeling important phenomena of biological interest once provided with the information on the initial state and reactivity parameters, or growing factors. Relationships existing between P systems and discrete linear systems are investigated. Finally, exploratory considerations are addressed about the possible use of P systems in characterizing the oscillatory behavior of biological regulatory networks described by metabolic graphs.