Problem with ADS em simulation for dualband filter

Hi guys, im trying to simulate a dual-band microstrip filter in ADS with the frequency at 2.4Ghz to 5Ghz.
I've already have all my components designed using tlines-microstrip and its under (jpeg1) and the graph obtained is under (jpeg2).
The problem occurs when i convert my filter into layout design(jpeg3) in ads.
The graph obtain through em simulation seems to be very weird and seems to be 1GHz shifted.(jpeg4)

For em simulation:
substrate fr4=4.7 dieelctric constant
height=1.6mm
frequency plan=1.4GHz to 7GHz, Npts=50

Ive tried changing the frequency plans , Npts and the frequency type to adaptive but nothing changes

Could you guys help me figure out the problem? Thank you

jpeg1

jpeg2

jpeg3

jpeg4

I doubt that the TL crosspoint model can be applied to the extreme stub widths in the actual implementation. The assumption of a 1-D wave transmission isn't met. The edges of the shunt stubs seem even to touch the series stubs.