I don't think many people noticed this story, but on Thursday Sourcefire Labs published A New Detection Framework on the VRT blog and a NRT page on their labs site. I had a small part in this development due to the Incident Detection Summit I organized late last year. Sourcefire sent an army of developers (I think they had the biggest contingent) to the conference and clearly enjoyed participating. During the event they spoke to participants from multiple security teams and had follow-up discussions with several of us.
One item we emphasized with Sourcefire was the need for analysis of file contents, not just network traffic. As Matt mentions in his latest post, Mike Cloppert and his team have used these approaches very effectively and have even published components of their work as open source projects like Vortex by Charles Smutz. In my NSM in products post last year I called this extracted content and listed it as one of the forms of NSM data.
What does this mean? The basic idea is that you extract content from network traffic, analyze it, record metadata, and so on, and then provide that information to a security analyst. That may sound like an anti-malware approach, but the idea is to provide indicators, not necessarily block transmission. In any case, Sourcefire published a presentation on their site on what their beta code can do. I'm really glad to see them working on this problem and sharing results in a form that interested parties can download and test.