We present an analysis of seven primary transit observations of the hot Neptune GJ436b at 3.6, 4.5 and $8~mu$m obtained with the Infrared Array Camera (IRAC) on the Spitzer Space Telescope. After correcting for systematic effects, we fitted the light
curves using the Markov Chain Monte Carlo technique. Combining these new data with the EPOXI, HST and ground-based $V, I, H$ and $K_s$ published observations, the range $0.5-10~mu$m can be covered. Due to the low level of activity of GJ436, the effect of starspots on the combination of transits at different epochs is negligible at the accuracy of the dataset. Representative climate models were calculated by using a three-dimensional, pseudo-spectral general circulation model with idealised thermal forcing. Simulated transit spectra of GJ436b were generated using line-by-line radiative transfer models including the opacities of the molecular species expected to be present in such a planetary atmosphere. A new, ab-initio calculated, linelist for hot ammonia has been used for the first time. The photometric data observed at multiple wavelengths can be interpreted with methane being the dominant absorption after molecular hydrogen, possibly with minor contributions from ammonia, water and other molecules. No clear evidence of carbon monoxide and dioxide is found from transit photometry. We discuss this result in the light of a recent paper where photochemical disequilibrium is hypothesised to interpret secondary transit photometric data. We show that the emission photometric data are not incompatible with the presence of abundant methane, but further spectroscopic data are desirable to confirm this scenario.
