Sequences from phylogenetically diverse taxa (see Additional file 1: Figure S2). Selected representative sequences are shown beneath the logo. Partial-width logo letters correspond to positions with alignment gaps (i.e. fewer contributing sequences). Note that teleost fish lack the ASXL2 TF ORFanalysed, both for ASXL1 and ASXL2 (Figs. 4 and 5). This motif is commonly followed by a serine (S) or a hydrophobic residue, and ?in the case of ASXL2 only ?by a proline (P) residue typically located four amino acids downstream. Additionally, in the case of ASXL2, the 40 amino acid region upstream of the EH[N/S]Y motif is well-conserved; in particular, a distinct LE[A/ E]G[E/Q] motif was observed at the beginning of this region (positions 113?17 in Fig. 5). We searched the eukaryotic linear motif (ELM) database [32] using the predicted human TF peptide sequences to determine whether any of the highly conserved regions might comprise known functional motifs. This analysisrevealed that the core EH[N/S]Y sequence matches the metazoan Host Cell Factor-1 (HCF-1) binding motif (HBM), which has the consensus [D/E]HxY [33, 34]. For both ASXL1 and ASXL2, this was the most significant database match for the entire TF peptide sequence (p = 5.1 ?10-5). The EH[N/S] Y motifs also coincide with distinct peaks in synonymous site conservation (Fig. 3 and Additional file 1: Figure S3) indicating that these sites are subject to particularly strong evolutionary constraints. A zero-frame amino acid sequence corresponding to +1 frame EH[N/S]Y is necessarily highly constrained ?for example, the first two positions can only be R/G and T/A respectively; yet all four amino acids were common atDinan et al. Biology Direct (2017) 12:Page 8 ofthese positions, for example human ASXL1 and ASXL2 have zero-frame RTQLL and GAQLQ respectively at this site, confirming that conservation of +1 frame EH[N/S]Y is not due to zero-frame coding constraints.Structural analyses of TF peptidesNo significant homology was found between the TF peptide sequences and tertiary structural domains within the InterProScan [35] or NCBI conserved domains [36] databases. We used the Predictor of Natural Disordered Regions (PONDR? algorithm to infer ordered and disordered segments in the full-length ASXL1 and ASXL2 proteins of humans (Additional file 1: Figure S5; left panel, top and bottom, respectively) and in the predicted frameshift products ASXL1-TF and ASXL2-TF (Additional file 1: Figure S5; right panel, top and bottom, respectively). The TF regions were inferred to be largely PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/8627573 PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/22316373 disordered (Additional file 1: Figure S5; right panel, highlighted sections). However, it is notable that the C-terminal EH[N/S]Y motifs are predicted to be found within locally ordered segments of the transframe products. Vitamin D2 Combined with the analyses of residue conservation discussed above, these data suggest that the TF peptides do not harbour functional tertiary structures, but they may exert a regulatory impact by means of conserved short linear motifs.Analysis of ribosome profiling datasetsRibosomal frameshifting into the TF ORF would result in a proportion of translating ribosomes terminating at the TF stop codon, leading to a step-wise decrease in ribosome density at this site that might be apparent in ribosome profiling datasets [5, 37]. Due to variability in read density as a result of library preparation biases (nuclease, ligation, reverse transcription, PCR, etc), ribosome profiling is unlikely to be sensitive enough.