What's stopping us determining structure for all proteins? We need
more money? We need more people? We need better technology?
Each unknown structure has a different story. Importantly, not all of the proteins have well-defined structures. But this hasn't historically been appreciated, so lots of effort has been spent structurally resolving regions that are not that ordered. But there are many other explanations for a protein's structure being unknown, most of which fit into two categories.
The putative function can be uninteresting. Most researchers only express and purify proteins that are potentially interesting. There is usually some genetic or pharmacological manipulation that implies a given protein is "interesting." Only the large scale structural genomics projects randomly solve structures, aiming for completion. Even then, these projects have to prioritize proteins that look feasible "on paper."
The protein can resist structure determination Lots of proteins are difficult to express and purify, for a multitude of reasons. For example, the table I just copied from the NorthEast Structural Genomic consortium summarizes their pipeline as follows:
Expression tested 24,479
Expressed (E > 0) 18,992
Expressed-Soluble (S > 0) 13,946
Expressed-Soluble (ES >= 9) 11,054
Purification attempted 8,691
Purified (>= 0.5 mg yield) 6,493
HSQC spectra recorded 2,853
"Good" HSQC spectra 1,476
HTP crystal screening 2,845
Crystal hit 1,480 1,672
X-Ray structures 608
NMR structures 496
Total structures 1,061
So using a combination of X-ray crystallography and solution-state NMR spectroscopy, they solved 1061 structures of 25924 clones, almost all of which were tested for expression. (http://nesg.org/statistics.html)
Crystallography is as automated as it can get right now. Protein NMR and cryoEM have their own niches, and will continue to fill out the gaps. But the science is difficult, and of the well-defined structures, there are still many that haven't been well-described.
(If we could really solve the protein folding problem, we wouldn't need to solve structures at all...)