We assess the effectiveness of a non-parametric bias model in generating mock halo catalogues for modified gravity (MG) cosmologies, relying on the distribution of dark matter from either MG or $\Lambda$CDM. We aim to generate halo catalogues that effectively capture the distinct impact of MG, ensuring high accuracy in both two- and three-point statistics for comprehensive analysis of large-scale structures. As part of this study we aim at investigating the inclusion of MG into non-local bias to directly map the tracers onto $\Lambda$CDM fields, which would save many computational costs. We employ the bias assignment method (BAM) to model halo distribution statistics by leveraging seven high-resolution COLA simulations of MG cosmologies. Taking into account cosmic-web dependencies when learning the bias relations, we design two experiments to map the MG effects: one utilising the consistent MG density fields and the other employing the benchmark $\Lambda$CDM density field. BAM generates MG halo catalogues from both calibrations experiments excelling in summary statistics, achieving a $\sim 1\%$ accuracy in the power spectrum across a wide range of $k$-modes, with only minimal differences well below 10\% at modes subject to cosmic variance, particularly below $k<0.07$ $h$Mpc$^{-1}$. The reduced bispectrum remains consistent with the reference catalogues within 10\% for the studied configuration. Our results demonstrate that a non-linear and non-local bias description can model the effects of MG starting from a $\Lambda$CDM dark matter field.
Comment: 15 pages, 10 figures, 2 tables