Have a look at the virulence factor database (VFDB)

There you can retrieve the virulence factors data set:

http://www.mgc.ac.cn/VFs/download.htm

Hope this helps

There are few data bases of virulent factors. Here is one database you might want to have a look;  LLNL Virulence Database http://mvirdb.llnl.gov/

If you are specifically interested in virulent factors in your data set. Then you can also combine few of the databases to make a ''larger custom database'' and blast against it, thereby increasing your virulent factor database pool. However please check for duplications between the databases. 

Regards

Thank you a lot for your answers. I already used MvirDB with a sequence similarity cutoff of 80%. But the problem is that there are no outsequences which might have a better hit. Therefore I'm looking for a way to specifically "fish" out virulence factors from NCBI after BLASTx

At the CGE they have a bunch of very easy to use tools including Virulence finder. Although it's not directly meant for metagenomic data you could see if you get any hits or use their database as Rounik suggested.

https://cge.cbs.dtu.dk/services/VirulenceFinder/

Good luck

It depends on how do you want to use the information. If you just want to report for potential virulent factors then maybe try a lower cutoff of maybe 70%.  Or one possibility is that, you can use your blastx annotations and the annotations from the MvirDB and cluster them according to their shared name. If some of your annotations forms larger clusters with the MvirDB then take those clusters as potential virulent factors. This can help you initially to look for the potential virulent factors. After this you might want to go to your sequences individually and access/decide their potential to be virulent. Keep in mind this might be a labor intensive process and a tricky one.