Phylostratigraphic tracking of cancer genes suggests a link to the emergence of multicellularity in metazoa

In addition, the TBLASTN searches (10-15 E-value cut-off) were done against substantial trace and EST archives of Porifera, Cyclostomata and Chondrichtyes (phylostrata 6 and 11, Figure 1) as complete annotated genomes are still lacking for these internodes. The higher threshold for the trace and EST archives was necessary because of the different data structure.

Using the obtained BLAST output and the MS SQL database management system in a series of queries we mapped human genes according to the evolutionary origin of their founder genes on the currently best supported phylogeny (Figure 1) [15-19]. Taken together, our choice of internodes depended on the availability of complete annotated genomes, reliability of phylogenetic relationships and on the importance of evolutionary transitions.

We retrieved cancer associated genes from several public resources [20-23]. The full list of genes and datasets used in the analysis is listed in Additional File 1: Table S1. We performed a series of overrepresentation analyses for various combinations of these datasets, in a way that the frequency of cancer associated genes in every phylostratum was compared to the frequency of cancer associated genes in the complete genome (expected frequency) [1,2]. Obtained deviations are shown by calculating log-odds ratios and their significance were tested by two-tailed hypergeometric tests [49] corrected for multiple comparisons via a false discovery rate at the 0.05 level [50]. The caretaker and gatekeeper classifications were taken from the Entrez section in the CancerGenes database [20] as this was currently the only large scale annotated dataset that allowed this type of grouping in a straightforward way. Note that the distinction between caretakers and gatekeepers is not definitive; some genes were listed in both of these categories as they act as both caretakers and gatekeepers. However, all of this information is taken into account and does not pose a problem for the statistical treatment in the overrepresentation analysis. For the analysis of caretakers and gatekeepers we used a more stringent subset of Entrez CancerGenes by taking into account only those that have definitive evidence that they are mutated in human tumours (Sanger Cosmic list [22]). However, similar results are also obtained when the other databases were used (data not shown).

EST: expressed sequence tag; NCBI: National Center for Biotechnology Information.

TDL initiated the study and carried out all the bioinformatic analysis. DT contributed to the data interpretation and further development of the study. TDL and DT wrote the manuscript.

TDL was supported by a Research Fund of Republic of Croatia (Grant 098-0982913-2832 to Ð Ugarkovic), Zaklada Adris, Unity through knowledge Fund (Grant No. 49), as well as a postdoctoral fellowship of the MPG. Computational resources were provided by CSTMB and RBI Zagreb. We thank Robert Bakarić, Vedran Dunjko and Martin Sebastian Šestak for their technical support and John F Baines and Alexander Pozhitkov for their discussion and comments on the manuscript.