The measurements have shown that the parton distribution functions
(PDFs) in nuclear collisions are modified with respect to the free
nucleon PDFs. These nuclear modifications can be determined similarly as
the proton PDFs via global analysis. However, the data included in the
present nPDF fits constrain well the modification of quarks at x > 0.01
but the uncertainties of gluon modification especially at x < 0.01 are
large. In this talk I will discuss how the gluon modification could be
constrained with the p+Pb collisions at the LHC and which x regions can
be probed with inclusive hadrons and photons at different rapidities
according to the NLO pQCD framework. I will also introduce our spatially
dependent nuclear PDFs and demonstrate their impact to the centrality
dependence of particle production in nuclear collisions. In addition, I
discuss briefly about the inconsistencies between the LHC data for
charged hadron production and the NLO pQCD calculation with the present
fragmentation function sets.