You can, but it is probably best to remove sparse genes (genes missing from a high percentage of taxa, say 50%, or 80%). Also, probably good to remove taxa with a high proportion of missing genes. I am not up to date with the latest literature, though, so I don't know the most accepted thresholds.

Alternatively, you can reconstruct trees on a gene by gene basis, then use some tree conciliation / supertree method to obtain the final tree.