Statistical analysis

The raw measurements of the MFs, for each of the 96 wells (Fig. ​(Fig.1c-d),1c-d), underwent a two-step process. First, all edges were discarded. This was due to the commonly known possible evaporation of medium in the outer wells of multi-well plates. These wells were filled with medium with cells as all the others, but they were not analyzed, so only 60 out of the 96 wells were analyzed. Secondly, the exact values of the measurements (taken with a precision of 0.1 μT) were rounded up to integer values coinciding with the actual measurements within ±3 μT and ± 2 μT for the DC and AC MFs, respectively. This made possible to group together wells exposed to similar fields. Although seeded completely, only 8 wells of the control plate (wells D5 through E8) at a homogeneous DC MF of 41.7 μT were assayed and included in the statistical analysis. In each experiment, the absolute colorimetric readings of the exposed and also of the control wells were normalized (i.e., divided) by the average of the 8 control wells, hence the average control was always identical to 1 and the proliferation of the exposed wells are reported relative to control. After finding significant differences with a one-way ANOVA analysis, Dunnet’s test was used to compare exposed against control wells, and the Tukey HSD multiple comparison test was used to evaluate differences within the same plate, but for different AC MF intensities. All the statistical analysis was performed with the STATISTICA software on the spectrometric raw data (see Additional file 1: MagneticFields&ProliferationAssaysRawData.xlsx). 3D plots of relative proliferation were prepared with the Python Matplotlib package [40].