On the usage of photon PDF by Syscalc

Dear developers

I have doubts how the photon PDF is handled in MG5_aMC_v2_3_3
I am calculating the photo-production of a spin-2 particle at LHC using the default RS model of MG.
I do:

define p = a
generate p p > y

The PDF I am trying are

NNPDF23_nlo_as_0119_qed
NNPDF23_nlo_as_0119_qed_mc
CT14qed_proton
MRST2004qed_proton

In the run card I tried as beam: 1=proton, 2=photon from proton.
If I use beam = 2 I get a huge cross section (larger than gluglu>y), what cannot be,
If intead I use beam = 1 I get a reasonable behaved cross section, well behaved with m_y

For example LHC14, to m_y = 1 TeV
1: Cross-section : 9.87e-09 +- 1.528e-11 pb
2: Cross-section : 4.444e+18 +- 1.65e+10 pb

Up to here I could guess that the correct is use beam = 1 (just proton) and a PDF with QED part.
However when I use the Syscalc to calculate the alpha_s and PDF uncertainties I get as result alwyays:

The Syscalc.dat I use is

==========================
# Central scale factors
scalefact:
1 2 0.5
# choice of correlation scheme between muF and muR
# set here to reproduce aMC@NLO order
#scalecorrelation:
0 3 6 1 4 7 2 5 8
# PDF sets and number of members (0 or none for all members)
PDF:
cteq6l1.LHgrid
NNPDF23_nlo_as_0119.LHgrid 1
NNPDF23_nlo_as_0119_qed_mc.LHgrid
NNPDF23_nlo_as_0119_qed.LHgrid
CT10nlo.LHgrid
PDF4LHC15_nlo_mc.LHgrid
PDF4LHC15_nlo_mc_pdfas.LHgrid
================================

scale fact cross-section :11 -nan
...
pdf reweighted cross-section :2 -nan

I explitly check that events are generated.
I am not sure what happens... Maybe I should specify something regarding photons to be in initial state in syscalc?
Could you advice me please?
thank you
Xanda