Plant virus diseases are often difficult to diagnose because viruses cannot be seen. Normally one begins by learning common virus symptoms and knowing common virus diseases. One can then compare symptoms and field patterns of diseases with what is known for specific diseases. Symptoms by themselves are notoriously unreliable. They're reasonably good for standard diseases in standard places, but the farther you get from what you know the less reliable they become.
Confirming a virus disease
The classical method for confirming a virus disease is to inoculate a diagnostic host. A diagnostic host is one which gives a characteristic symptom for the virus in question and which distinguishes the virus from other viruses with which it might be confused. Sometimes one might inoculate several different hosts to be sure. The disadvantage of hosts is that the response takes days to weeks and considerably longer if you don't have the right plants at the proper age to begin with. Physical techniques for identifying viruses are much quicker, but usually more expensive. Electron microscopy is a quick technique, but electron microscopes are expensive and usually not readily available. More commonly one uses serology (antibodies specific to the virus). ELISA (enzyme linked immunosorbent assay) is the most common assay. The most common protocol (double antibody sandwich) is to adsorb the specific antibody to the well of a plate, "capture" the virus in the sample with this antibody and then detect the virus with antibody coupled to an enzyme. The most commonly used enzyme is alkaline phosphatase which hyrdrolyzes p-nitrophenylphosphate to p-nitrophenol which is yellow. Since each enzyme molecule hydrolyzes thousands of substrate molecules the response is amplified. ELISA assays require meticulous attention to detail to prevent nonspecific reactions which can be confused with a positive virus test. Often samples are sent to commercial companies which routinely perform ELISA.
Diagnosing unknowns
Diagnosis becomes more complex if one is not sure there is a virus and if one doesn't know what the virus is likely to be. To determine if there is a virus, inoculation to the same host as the sample is a good place to start. Inoculation to "universal" hosts such as Nicotiana benthamiana or Chenopodium quinoa (see Lost Crops of the Incas, National Academy Press) are also common techniques (there is no such thing as a truly universal host). Electron microscopy is useful because viruses have distinctive sizes and shapes. Extraction and analysis of dsRNA which is not found in healthy plants, but which is produced by most viruses is also useful. I commonly use a procedure called " minipurification" where the virus is partially purified by ultracentrifugation and then proteins are analyzed by SDS gel electrophoresis. This procedure works because viruses are larger than most cellular components and readily pellet in the ultracentrifuge. The virus capsid protein is often prominent among proteins in the ultracentrifuge pellet (see examples). The presence of an "extra" protein is an indication of a virus and the size of the protein provides useful information on what the virus might be.