I usually measure the average MW of a protein by injecting it neat onto a PLRP column using RP-HPLC. Then I deconvolute the MS. Would a similar method work on DNA that is about 6000 Da?
It's all about the calibration. Calibrate the instrument before and after the measurement with knowns that bracket the m/z range of interest. If you have an overlapping isotopic series you will severely degrade the exact mass determination. If you are doing charge state deconvolution, you lose accuracy roughly by the multiple of the charge state that is being deconvolved (e.g., a 16+ charge state with an delta m/z error of 0.001 would be a 0.016 Da error on the zero charge mass. You can improve that roughly by the sqrt of n, where n is the number of isotopic masses and charge states that are included in the deconvolution. (Standard Student's t-text applies to the confidence intervals), assuming each is deconvolved to zero charge separately. You also need to assess your isotopic patterns for overlaps, if you use them. see: Method Isotopic Vector Angle Analysis in Mass Spectrometry v2
Just out of interest: are you planning to run negative ion mode ESI-MS for your DNA analysis? And do you plan to use any kind of ion paring agent during reversed phase analysis (other than that used for molecular weight determination of proteins)? From experience, I know that measurement conditions greatly impact the sensitivity and accuracy of DNA analyses.
What range of charge states are you expecting? In other words, do you think that deconvolution is required for your DNA sample of interest? And what kind of mass accuracy would you ideally like to obtain for your 'average MW'? It it simply to distinguish your product from an n-1 contamination, or are you analyzing something else?
It MS strikes me as a very poor way to determine molecular weight for such a large and possibly fragile molecule. The common solution based ways using GPC (molecular size separation) and light scattering would seem to be much more analytically robust..