In information theory, a signature is characterized by the information content as well as noise statistics of the communication channel. Biosignatures have analogous properties. A biosignature can be associated with a particular attribute of a pathogen or a host. However, the signature may be lost in backgrounds of similar or even identical signals from other sources. In this paper, we highlight statistical and signal processing challenges associated with identifying good biosignatures for pathogens in host and other environments. In some cases it may be possible to identify useful signatures of pathogens through indirect but amplified signals from the host. Discovery of these signatures requires new approaches to modeling and data interpretation. For environmental biosignal collections, it is possible to use signal processing techniques from other applications (e.g., synthetic aperture radar) to track the natural progression of microbes over large areas. We also present a computer-assisted approach to identify unique nucleic-acid based microbial signatures. Finally, an understanding of host-pathogen interactions will result in better detectors as well as opportunities in vaccines and therapeutics.