We present the analysis of broad absorption lines (BALs) seen in the spectrum of the zem≃4.92 QSO SDSS J160501.21-011220.0. Our high spectral resolution UVES spectrum shows two well-detached absorption line systems at zabs=4.685 and 4.855. The system at zabs=4.855 covers the background source completely, suggesting that the gas is located outside the broad emission line region. On the contrary, the system at zabs=4.685, which occults only the continuum source, has a covering factor of the order of 0.9. Physical conditions are investigated in the BAL system at zabs=4.855 using detailed photoionization models. The observed Hi absorption line together with the limits on CII and SiII absorptions suggest that 16<logN(HI)(cm–2)<17 in this system. Comparison with models show that the observed column densities of NV , SiIV and CIV in this system require that nitrogen is underabundant by more than a factor of 3 compared to silicon if the ionizing radiation is similar to a typical QSO spectrum. This is contrary to what is usually derived for the emission line gas in QSOs. We show that the relative suppression in the Nv column density can be explained for Solar abundance ratios or abundance ratios typical of Starburst abundances if an ionizing spectrum devoid of X-rays is used instead. Thus, if the composition of BAL is like that of the emission line regions it is most likely that the cloud sees a spectrum devoid of X-rays similar to what we observe from this QSO. This is consistent with the fact that none of our models have high Compton optical depths to remove X-rays from the QSO. Similar arguments lead to the conclusion that the system at zabs = 4.685 as well is not Compton thick. Using simple Eddington arguments we show that the mass of the central black hole is ∼7.1x108M☉. This suggests that the accretion onto a seed black hole must have started as early as z∼11.