Under "advanced search", one can enter a molecular weight range. You'll probably get a lot of hits. If the compound is unknown (such as from a natural product), consider possibilities such as loss of water, ammonia, carbon dioxide in your search. Also consider charge carriers such as potassium and sodium. In the link below, the extracted capsaicin compounds exhibited [M+Na]+ as the predominant compounds.
Assuming natural products, also do a literature search on the organism from where the extract came from. I find that with plants, searching at the family level (as well as the genus and species) would suggest compounds that have already been discovered.
ChemSpider's advanced search may be of use. If you know the molecular weight you probably have a mass spectrum, correct? There are some unmistakable mass fragments that pop up for mass spec (such as ~91, which indicates a phenyl or benzyl ring). You could draw a structural fragment in your advanced search in addition to searching within a mass range.
Also you have obviously extracted your product, which means you know its solubility characteristics, which is also a search option in advanced search.
You could try this site > http://sdbs.db.aist.go.jp/. It's not specific just for antibiotic though. And I think you need to add more data such as IR spectra, H and C-NMR for better understanding of your compound.
In the supplementary material of that article their is a large list f DBs : https://www.researchgate.net/publication/267467312_Current_approaches_and_challenges_for_the_metabolite_profiling_of_complex_natural_extracts
Article Current approaches and challenges for the metabolite profili...
The commercial search engine SciFinder is a very good place to look. It is quite expensive though. If you want to find if your molecule is novel, you'll have to search more than antibiotic databases.
Is this a pure compound? Even small amounts of impurities with high molar absorptivities can completely fool you. For instance, if your compound has a molar absorptivity of 100 M-1cm-1 and you have 0.1% impurities with 10^5 M-1cm-1 (not an outrageous number at all, particularly at low wavelength) then your impurities will account for half of your signal!
What are you even doing? It is very hard to identify a compound by UV-vis if we have no idea what types of reactions you are doing, what the reactants are, what the reaction conditions are, etc...
If you have a few guess based on what kinds of products you would expect and these are stable compounds, then I would suggest that you acquire spectra of the possibilities in the same solvent (if water, at the same pH) and see if you can get things to match up. If you can't get these compounds then you really need an NMR and a mass would be helpful.