Searching for drugs in equine urine is not easy unless one knows in advance which drugs to look for. With mass spectrometers we have to search for specified spectra or specific ion responses. Our eight GC mass specs do lots of screening for drugs and in an average year they would record about 24,000,000 spectra in the search for these drugs. But the dastardly villains of the racecourse have been known to unleash drugs on the sport for which we weren't searching - fiendish swine. We don't have the time to look at all those spectra and even if we did how would we know what is significant?

To rephrase the question - how do you find a needle in a haystack if you don't know what a needle looks like and you don't have much time?

We started by having a hard look at our spectral heap. We run a variety of overnight screens which search for different groups of drugs. In each screen the same extraction and derivatisation procedure is applied for each set of samples. Almost all of these spectra come from compounds produced in these processes from the extracted urine. Consequently the same spectra tend to recur in each sample screened. Now everyone knows how good computer library systems are at identifying a spectrum it has seen before but how about asking a library system to look for spectra the computer hadn't seen before?

For our first trials we used a screen for opiates which is run on the Magnum Ion Trap (Finnigan) which scans in full scan mode using positive chemical ionisation. At the last ANZSMS conference we described (possibly not in the most glowing terms) the use of this system for drug detection. Since then this mass spec has proved to be an excellent screening machine provided we fed it with what were to us quite clean samples!

Using the Finnigan MS Procedure Language we wrote a programme for the data system (a 486DX/33 PC) which looked at the spectra of all GC peaks from the overnight runs. We also produced an augmented spectral library which remembered scan number, the number of times each spectrum was top match and the run details of the last four GC runs to access each spectrum. Spectra found in this library which match on scan number, fit, purity and reverse fit are ignored while those not recognised are placed on a printed list to be later investigated by hand. This library now has about 200 compounds in it although, of course, it doesn't have the foggiest what the compounds are!

So far the system has proved to be worth support. It detects most random quality control samples and has found several very interesting spectra which have lead to further investigation. As could be expected it is not as good as the specific drug detection system for specified drugs. Difficulties arise when GC retention times change as the library system relies a lot on scan numbers. It also relies on consistency in the extraction and derivatisation methods, mass ranges and scan speeds.

However, as we old gold miners say, there is a touch of colour in the pan and so we look forward to a time when dastardly villains of the racecourse get their comeuppance - keep an eye on the racing pages for future success!