Amyloid beta (A beta) peptides are considered the major causative agents of Alzheimer's disease (AD). In a widely accepted mechanism for AD pathogenesis, A beta peptides are proposed to play multiple roles in damaging brain cells and their synaptic communications. Due to the heterogeneous nature of A beta oligomers, their in vivo structures have not been understood. Most experimental and computational studies favored beta-rich structures of A beta as observed in A beta fibrils. In this in silico study, we investigated an alternative perspective on the structures and function of A beta oligomers in the cell membrane. Transmembrane alpha-helix bundles of the A beta(17-42) tetramer and trimer were observed in extensive temperature replica exchange molecular dynamics (REMD) simulations. We observed three minima on the free-energy landscape of each oligomer, namely, A, B, and C for the tetramer and D, E, and F for the trimer. Except for F, the minima consist of 4 or 3 parallel helices spanning across the membrane model dipalmitoylphosphatidylcholine. Replica exchange molecular dynamics-umbrella sampling (REMD-US) simulation was applied to study the process of a Ca2+ crossing the pore formed by the alpha-helix bundles in A-E in comparison to that in a calcium channel and a proton channel. REMD-US reveals that A, C, and D allow Ca2+ to cross their pore with a free-energy barrier comparable to that found for the calcium channel. In contrast, the free-energy barrier of a Ca2+ ion crossing B, E, and the proton channel is significantly higher. This result suggests that A beta peptide oligomers could form transmembrane alpha-helix bundles that provide feasible pathways for Ca2+ transport. This is an intriguing observation that will stimulate further studies.