Aims. We studied the multi-frequency predictions of various annihilating dark matter (DM) scenarios in order to explore the possibility to interpret the still unknown origin of non-thermal phenomena in galaxy clusters. Methods. We consider three different DM models with light (9 GeV), intermediate (60 GeV), and high (500 GeV) neutralino mass and study their physical effects in the atmosphere of the Coma cluster. The secondary particles created in the neutralino annihilation processes produce a multi-frequency spectral energy distribution (SED) of non-thermal radiation and also heat the intracluster gas, which we test against the observations available for the Coma cluster from radio to gamma-rays. The various DM-produced SEDs are normalized by the condition to fit the Coma radio halo spectrum, thus obtaining best-fit values of the annihilation cross-section σV and of the central magnetic field B 0 . Results. We find that it is not possible to interpret all the non-thermal phenomena observed in galaxy clusters in terms of DM annihilation. The light-mass DM model with 9 GeV mass produces too little power at all other frequencies, while the high-mass DM model with 500 GeV produces a large excess power at all other frequencies. The intermediate-mass DM model with 60 GeV and τ ± composition is marginally consistent with the HXR and gamma-ray observations, but narrowly fails to reproduce the EUV and soft X-ray observations. The intermediate-mass DM model with 60 GeV and bb composition is, on the other hand, always below the observed fluxes. We note that the radio halo spectrum of Coma is well fitted only in the bb or light-and intermediate-mass DM models. We also find that the heating produced by the DM annihilation in the centre of the Coma cluster is always larger than the intracluster gas cooling rate for an NFW DM density profile and it is substantially smaller than the cooling rate only for a cored DM density profile in light-mass DM model with 9 GeV. Conclusions. The possibility of interpreting the origin of non-thermal phenomena in galaxy clusters with DM annihilation scenarios requires a low neutralino mass and a cored DM density profile. If we then consider the multi-messenger constraints to the neutralino annihilation cross-section, it turns out that this scenario would also be excluded unless we introduce a substantial boost factor that represents DM substructures. If we relax the condition to fit the Coma radio halo and we consider the EUV and HXR detections as upper limits for the non-thermal emission, together with the gamma-ray limit, then the limits on σV are less stringent than those obtained by the multi-messenger analysis.