The Molecules Gateway lists 150,777 different entries. Of these, 1,031 are present in the seven different unfermented media used to cultivate the strains and derive from the complex ingredients (i.e. soy peptone, soluble starch, casein hydrolysate, yeast extract, meat extract, soybean meal and bacto-peptone ) used for media preparation. These molecules are labeled as such in the Molecules Gateway.
Annotation levels and annotation tools
As explained here, the annotations derive from a decision tree and are classified on the basis of their likely reliability from 1 (least reliable) to 4 stars (most reliable). In addition, a small number of molecules has received a 5-star score because they were identified using reference standards or manual curation. Finally, entries without any predicted molecule (0 stars) consist of two subgroups, depending on whether or not there is a molecular formula predicted by SIRIUS (0_MF). The distribution of molecules by annotation level can be seen below.
The three annotation tools – Compound Discoverer (CD), MolDiscovery (MD) and MS2Query (MQ) – predicted molecules at very different rates, ranging from almost the entire portion for CD to just 12% for MD. Of note, molecule prediction by a tool does not imply that the prediction is correct.
Frequency of molecules
Frequently occurring molecules are expected to represent medium components, molecules from primary metabolism or common specialized metabolites. Most molecules are present in a few extracts only, and only 3,120 molecules are contained in more than 200 extracts. See the frequency of molecules present in the 1–200 extract range.
Taxonomic origin
Molecular diversity
How different are the molecules listed? This question can be answered, by looking at the chemical relatedness and originating biosynthetic pathway for the 5,660 unique InChIKeys listed in the Molecules Gateway (1417 molecules arranged into families and 4243 molecules forming single nodes), and at the distribution of exact mass and retention time for the 58,093 molecules with 1 through 5 annotation confidence level. These analyses indicate that all major biosynthetic pathways are represented, that a limited number of closely related molecular families occurs and that there is no obvious bias in retention time or molecular weight.
