The attached plot shows the Vy:Vx distributions in the STAR global coordinates, center is around (0.05, -0.35). If I translate this to the PXL local coordinate (dominant translation is the tpc2global). The Vx, Vy position are 0.27 and -0.25 cm. This looks to be consistent with the hit density distribution seen from the online plot.

The red lines indicate the PXL center position in the global coordinates. Black lines are the fitted means to the vertex distributions in data for fabs(Vz)<10cm.

So to make the beam more centered w.r.t to the PXL, we want to move the beam orbit to negative x direction (which is to north) by  0.28 cm and up by about 0.21cm.