Accidental eccentricity alone will rarely produce net tension at an interior ground-floor column in a genuinely regular frame — it shifts the storey shear distribution, but the overturning demand it adds at an interior line is small because the interior column sits near the centre of rigidity. Before you chase the eccentricity, export the column axial force for each individual load case rather than the combination, and check the sign of the EQx component on its own. If EQx alone produces an axial force larger than your dead load reaction, the frame is transferring far more overturning through that column line than a regular frame should, and that points at a stiffness anomaly rather than a load anomaly.
The usual culprit in this exact situation is an unintended outrigger. Check whether any beam framing into that column has been assigned a much larger section than its neighbours, or whether a beam has been modelled with end releases on one side and not the other. A single fixed-fixed deep beam in a frame where everything else is shallower will couple the two adjacent columns into a vertical truss and pump axial force into them under lateral load. Also check for accidentally duplicated members at that joint — two coincident beams double the stiffness and the axial coupling with it, and the model will look perfectly normal in plan.
