It can be done in different ways. If you have don't explicitly tell TOPAS what to do it will default to 1/SigmaYobs2 weighting where weights are derived from experimental uncertainties. For a .xye file SigmaYobs would be read from the data. For a .xy file Poisson statistics of SigmaYobs = Yobs0.5 would be assumed. If you search topas.inc for keyword "weighting" you'll see macros for e.g. TOF data or other choices. The basic equation will be something like:
weighting = If(SigmaYobs < 1, 1, 1/SigmaYobs^2);
The "if < 1" is to catch erroneously small values of esds (e.g. 0.0) which may be mistakes in the data file and would lead to silly weights. Depending on the scale of your yobs this might need changing.
If you put your own weighting equation in the INP file it will replace the default.