The search engines are very easy to use. In your mass spectrometer software, first select spectrum list. Then copy the spectral listing into WordPad or Notepad and save as a plain text (WordPad) or ANSI text (Notepad) file. Repeat for additional data. These files are then zipped together and uploaded to the Sun Grid. The search is then initiated.

The procedure for using FragSearch and IndexSearch on your data is summarized in this simple four step procedure.Two videos (RealPlayer format) are also available. For a brief (168 seconds) introduction to the Sun Grid, please see Introduction to the Sun Grid. For a brief (298 seconds) demonstration on using FragSearch and IndexSearch please see Using FragSearch and IndexSearch

Four examples of results from IndexSearch are described below: (download a few of these examples to try it yourself on the Sun Grid)

Xemilofiban is an example of a compound that has a common formula (C18H22N4O4 free base form). There are about 700 compounds in PubChem with that same molecular formula. In addition there were 938 compounds within 10 ppm of this experimentally determined exact mass. By parallel processing the search, the results are typically obtained in less than 14 minutes elapsed time, although about 80 minutes of CPU time is used. The search time increases with molecular weight, number of fragments, and number of compounds with the same molecular formula.

Note here that xemilofiban (PubChem ID 3033830) is found even though its exact mass includes the hydrogen chloride. Indexing allows us to index common salts as their free bases or conjugate acids, the mass that would be observed by ESI. Note also that the very closely related compound (PubChem ID 9946860) is easily distinguished by the search engine.

To explain the results output file further:

If a compound has three asterisks added after the ID number, it is considered "common" (e.g a compound in commerce)

The partition score is based on mathematical partitioning, and overall results are sorted by this score; the numbers are only relative to other compounds in that solution set.

The fragment score is what percentage of the fragments in the data can be explained by breaking up to four bonds of the compound simultaneously.

The search time increases with molecular weight, number of fragments, and number of compounds with the same molecular formula.

Malathion illustrates that quaternary compounds can be found by the search engines. The phosphanium salt (PubChem ID 4712968) possibility is found here; it is consistent with all of the data.

CID 4004 gets three asterisks from the program as it is a common compound, malathion. There were 781 compounds within 11 ppm of this experimental molecular weight; this search was done in less than five minutes.

Dimethyldioctadecylammonium chloride or bromide is a very recent example of a quaternary compound identified in an article in JASMS May 2008. A resolving power of 950000 is not needed if you have IndexSearch available. The elapsed time for this search of the ten million compounds in our PubChem index was two minutes and 15 seconds. (The isotope data used for this IndexSearch analysis was computer generated, because this data is not given in the paper.)

Fluconazole is a example of a compound that does not get the highest score. In this case one compound, closely related to fluconazole, outscored the correct structure (PubChem ID 3365). There were 2259 compounds in the index with a molecular weight within 12 ppm of this experimental data; this search was done in less than five minutes. Only CID 3365 is a "common" compound found in this result set; it was labelled with three asterisks.