Indeed, native or reducing PAGE and SEC-MALLS would be the first choices in identifying the MW, since they are practical and easy to interpret.

Depending on the expected MW and charge accumulation on the protein surface (expected charge variant profile) intact mass analysis can be performed on any HRMS system including QTOF. Deconvulated spectra can indicate the MW of the targeted collagen but also the PTM profile (not site specifically). LC-MSMS is not an option due to its application being more amenable to quantifying signature peptides and getting results in terms of quantitative proteomics. It is more sensitive than HRMS systems but not accurate enough in terms of exact mass monitoring. In QTOF data you may see many isotopic variants, charge envelope profiles, and possibly in-vivo PTM variants for target proteins. Therefore it is a bit overwhelming to identify MW by only analyzing the raw spectrum. Therefore software automatically deconvolutes the spectrum would be helpful. If the sample is a kind of protein mixture intact mass analysis (do not confuse it with top-down analysis, we do not fragment the precursor in this acquisition mode) is nearly impossible to get exact masses, even if the resolution is perfect. Very effective chromatography prior to analysis or injecting purified samples into the system (fractionated for instance) must be taken into account.

MALDI-TOF on the other hand would be more effective because MALDI ionization does not let compose multiply charged variants. ESI in LC-MS is prone to derivate a high number of charged proteins, thus MALDI produces reduced variants and presents intact mass reliably. Do not forget to consider MW range (predicted) of the collagens high order structures would be more complicated and mass ranges of the mass filters of the instrument must be capable of detecting the multiply charged species of the target protein. In the case of extended MW ranges, I mean analyzing collagen peptides and proteins simultaneously, MW fractionation using cut-off filters and subsequent MS analysis of the fractions should be the choice. Otherwise, you may both lose sensitivity and get complicated masses...

Garbage in garbage out...!