OrthoList: a compendium of C. elegans genes with human orthologs

Abstract

Background: C. elegans is an important model for genetic studies relevant to human biology and disease. We sought to assess the orthology between C. elegans and human genes to understand better the relationship between their genomes and to generate a compelling list of candidates to streamline RNAi-based screens in this model.

Results: We performed a meta-analysis of results from four orthology prediction programs and generated a compendium, "OrthoList", containing 7,663 C. elegans protein-coding genes. Various assessments indicate that OrthoList has extensive coverage with low false-positive and false-negative rates. Part of this evaluation examined the conservation of components of the receptor tyrosine kinase, Notch, Wnt, TGF-ß and insulin signaling pathways, and led us to update compendia of conserved C. elegans kinases, nuclear hormone receptors, F-box proteins, and transcription factors. Comparison with two published genome-wide RNAi screens indicated that virtually all of the conserved hits would have been obtained had just the OrthoList set (∼38% of the genome) been targeted. We compiled Ortholist by InterPro domains and Gene Ontology annotation, making it easy to identify C. elegans orthologs of human disease genes for potential functional analysis.

Conclusions: We anticipate that OrthoList will be of considerable utility to C. elegans researchers for streamlining RNAi screens, by focusing on genes with apparent human orthologs, thus reducing screening effort by ∼60%. Moreover, we find that OrthoList provides a useful basis for annotating orthology and reveals more C. elegans orthologs of human genes in various functional groups, such as transcription factors, than previously described.

Figure 1. Comparison of four orthology prediction programs queried for C. elegans orthologs of human proteins.

This diagram is modified from VENNY (see Materials and Methods). Each program is named above the oval representing its results, with the number of C. elegans orthologs and in-paralogs found by the program shown. The table gives an overall measure of how many genes were found by one or more programs (regardless of which one(s) found them). The numbers in the overlapping and non-overlapping areas of the Venn diagram indicate how many genes were found by overlapping or unique sets of programs. The font size used for these numbers indicate how many programs that number of genes was found by: numbers corresponding to genes found by a single program are shown smallest, whereas the largest font denotes the number of genes found by all programs. The data underlying this diagram can be seen in Table S1. A measure of the similarity and divergence between programs can be found in Table S2.

Figure 2. Examining OrthoList specificity and sensitivity.

Venn diagrams comparing C. elegans gene families, and their previously defined conserved subsets, to members of the same families found in OrthoList (see Materials and Methods). For each family the overlap between OrthoList and the previously defined conserved subset is shown (the percentage refers to how well covered by OrthoList the conserved subset is). The possible homologs missing from OrthoList (putative false-negatives, shown above each overlap) and those found in OrthoList not previously defined as homologs (putative false-positives, shown below each overlap) are based on homology assignments of the original compendia for each of these families (see Materials and Methods, and –[28], [56]). As discussed in the main text, the number of false-negatives and false-positives may actually be lower. A) Kinases (see Table S3C for source data). B) NHRs (see Table S4A for source data). C) F-box proteins (see Table S4B for source data).

Figure 3. OrthoList coverage of conserved signaling pathways.

Genes in bold are found by at least one orthology-predicting program, and thus included in OrthoList. The source data for this figure can be found in Table S5. A) RTK/Ras/MAPK pathway (reviewed in [34]). Note that ras-1 and ras-2 have not been defined functionally, although they are highly conserved. B) Notch pathway (reviewed in reviewed in [32]). C) TGF-ß pathway (reviewed in [33]). We note that tag-68, was only defined by conservation and no phenotype has been associated with its loss. D) Wnt pathway (reviewed in [30]). Note that our analysis was restricted to the conserved, canonical Wnt pathway. E) Insulin pathway. We specifically highlight the six insulins (daf-28, ins-1, ins-4, ins-6, ins-7 and ins-8), out of forty, that have been found (by overexpression, biochemical methods, RNAi or by existence of a semi-dominant allele) to be functional (reviewed in [31]).

Figure 4. OrthoList coverage of hits from genome-wide RNAi screens.

Venn diagrams analyzing hits obtained from RNAi screens examining (A) cell division and (B) endocytic/secretory trafficking . For each screen the overlap between OrthoList and the orthologous subset of hits is shown (percentage refers to how well covered by OrthoList this conserved subset is). Orthology assignments for hits missing from OrthoList (shown above overlap) and those found in OrthoList that were not called homologs in the original publications (shown below overlap) were confirmed by TreeFam and/or RBH (see Materials and Methods). Source data for these diagrams can be found in Tables S6 and S8.

Figure 5. OrthoList coverage of a transcription factor compendium.

A) Venn diagram comparing the wTF2.0 compendium to OrthoList. Source data for this diagram is found in Tables S9A, B. We find that OrthoList contains ∼98% of TFs previously scored as having human orthologs (overlap). In addition, OrthoList contains 182 TFs not scored in wTF2.0 as having orthologs. B) Distribution of TFs found to have orthologs by OrthoList, but not by wTF2.0. Source data for this table is found in Tables S9C–E.