Table 1. Statistics of orthologous proteins of B. belcheri and B. floridae.

	R1	R2	R3
Ortholog of the given B. belcheri protein
is present in the B. floridae proteome	93	93	93
is absent in the B. floridae proteome	7	7	7
Ortholog-pairs of B. belcheri and in B. floridae
have the same Domain Architecture	53	44	51
have different Domain Architecture	40	49	42
Ortholog-pairs of B. belcheri and in B. floridae*
have the same Domain Architecture*	93	93	93
have different Domain Architecture*	0	0	0
are present in the Swiss-Prot database	74	74	74
are absent in the Swiss-Prot database	23	23	23
Orthologs in Swiss-Prot have Domain Architectures
identical with those of
both B. belcheri and B. floridae	29	23	30
B. belcheri only	15	14	17
B. floridae only	8	15	7
neither B. belcheri nor B. floridae	22	22	18

One hundred proteins of B. belcheri, containing at least two Pfam-A domains, were selected from dataset Branchiostoma.belcheri_HapV2_proteins.fa (release 1, R1) and their equivalents were identified in dataset Branchiostoma.belcheri_v15h11.r2_protein.fa (release 2, R2) and dataset Branchiostoma.belcheri_v18h27.r3_ref_protein.fa (release 3, R3). Orthologs of the selected proteins were identified by the reciprocal best-hit method, using the B. floridae section of NCBI’s non-redundant protein sequence database and UniProt KB’s Swiss-Prot database. The domain architectures of the orthologs (defined as the linear sequence of Pfam-A domains) were compared as described in the main text. The domain architectures of B. belcheri and B. floridae proteins are compared in Supplementary Tables 1 and 2, the domain architectures of the B. belcheri proteins from the three different releases are compared in Supplementary Table 3.

*The lines marked with an asterisk refer to analyses of protein sequences corrected by FixPred. The observation that FixPred correction eliminates domain architecture differences of orthologs of B. belcheri and B. floridae indicates that these differences were due to errors of gene prediction.