Existing data sets annotating toxins include proteins and toxic peptides. They include bacterial, animal, and plant toxins (classified by source organism). Often, they other proteins associated with  toxins , even though they do not have  toxic effect. Here, we present a manually curated data set of bacterial proteins acting as toxins, called exotoxins, along with a curated data set of animal toxins and two control sets for non-toxic secreted proteins from animals and bacteria. Toxins in our data set were selected based on reports of secretion, and their ability to directly target and/or disrupt molecular mechanisms essential for the survival of the host organism. We compared simple biophysical features such as protein length, isoelectric point and amino acid composition to their secreted non-toxic counterpart, and between the kingdoms. While both animal and bacterial toxins use more cysteines than their control, animal had an additional over-representation in cysteines. In addition, we found animal toxins to be substantially shorter on average than bacterial toxins and had their isoelectric point shifted more toward acidic pH values. Overall, the differences in biophysical features between animal and bacterial toxins suggest to keep the two sets distinct.

Data to: Jiménez Soto, Luisa Fernanda, SSTDBB: Do You Speak Toxin? The Message of Bacteriophages; Subproject Translation. Research Report. ReNaTe (2026-03-12) https://doi.org/10.34657/31485