2.1. The majority of A. thaliana defensins belongs to class I defensins

In order to identify class II defensin sequences, we designed a semiautomatic pipeline (Fig. 1). For that, initially all proteins from the A. thaliana Uniprot database were downloaded. The dataset consisted of 86,486 sequences (March 2017). From this dataset, 387 sequences were retrieved by using regular expression (RegEx) search (step 2, Fig. 1). From these, 285 had up to 130 amino acid residues (step 3, Fig. 1). This criterion allows eventual larger C-terminal prodomains to be identified. Then, we used a PERL script to select the sequences with the following flags: hypothetical, unknown, unnamed and/or uncharacterized (step 4, Fig. 1), resulting in 15 sequences. From 15 sequences, seven were incomplete and therefore were discarded, (step 5, Fig. 1). From the remaining sequences, two sequences without signal peptide or with transmembrane domains were discarded (step 6, Fig. 1). Finally, from six remaining sequences, two sequences with a potential C-Terminal prodomain were selected, with accession codes A7 REG2 and A7 REG4 .