Star-forming main sequence of galaxies selected in this work in the redshift range 4 < z < 6. The best-fit relations are derived in the mass range where we have 90% mass completeness. For comparison, the high redshift and low-mass extrapolation of the MS relations by Schreiber et al. (2015) and Speagle et al. (2014) (obtained at lower redshifts and higher M_⋆) are shown with dashed orange and dotted fuchsia lines, respectively. Bottom panel shows the same as the top, but for the redshift range 6 < z < 10. The Bayesian best-fit relation obtained in the lower redshift bin in included as a pink continuous line for comparison.