Improve gene name matching

[gaurav]

Some genes (e.g. "SET", "KIT") are being matched to the English words (i.e. "set" and "kit") rather than to the gene. We need to improve this somewhat. Some possibilities:

• Can we ensure that the text is treated case-sensitively? I suspect it's being matching in a case-insensitive fashion right now.
• Can we use a library or dataset to look up gene names in the text separately from the SciGraph extraction? We'll get more false positives, but this might be useful.