Computational protocol: Detection of COPB2 as a KRAS synthetic lethal partner through integration of functional genomics screens

Similar protocols

Protocol publication

[…] Consider Figure . In the third step, the three selected screens are aggregated. In order to obtain such an aggregated gene ranking that is maximally consistent between the screens, we re-rank the genes such that the deviation of the aggregated ranking is minimal with respect to the three original rankings. Here, we use Spearman's footrule to quantify differences between rankings.Spearman's footrule calculates the difference between a set of ranked lists by measuring the total number of displacements for an element (here gene) across the lists. The RankAggreg algorithm aims at minimizing Spearman's footrule between the input lists and the output aggregated list []. Please see Methods for more details on Spearman's footrule distance. [...] The above analysis shows that standard method, RSA, and RIGER are not inconsistent and that there is not one method that can provide the one and only true ranking. At the same time, they are not in perfect agreement. To reflect this situation we consider the rank aggregation for all three screens as described above together with RIGER and RSA for the two screens of Luo and Wang. This is summarized as step 4 in Figure .Table shows the top three genes for this integrated ranking: coatomer protein complex, subunit beta 2 (COPB2), sprouty homolog 1, antagonist of FGF signaling (SPRY1), and nuclear receptor corepressor 1 (NCOR1). These three genes ranked consistently among the top 10% of all RankAggreg runs, with COPB2 ranking 98 times, NCOR1 100 and SPRY1 88 times among the top 10%. Further, these genes were the only ones among the top 10% ranks of RankAggreg that also ranked high according to RIGER and RSA. The functionalities of all three genes were explored and associated with the nine ’hallmarks of cancer’: the nine properties that are shared among cancer cells, and are necessary for tumor initiation and expansion []. The function of NCOR1 relates to the ’Resisting Cell Death’ property. The functions of SPRY1 and COPB2 are associated with the ’Sustaining Proliferative Signaling’ property. Information on the genes’ function was collected from GeneCards and Pubmed (www.ncbi.nlm.nih.gov/pubmed/) and Gene Cards (www.genecards.org) []. Silencing of SPRY1 has been found to trigger complete regression of RAS mutant cells in the human childhood rhabdomyosarcoma (RMS) []. COPB2 ranked best when the cumulative rank across all methods was considered, thus we focused on COPB2 in our experimental validations.We examined the expression levels of KRAS, COPB2, NCOR1 and SPRY1 across the five colorectal adenocarcinoma cell lines available in TCGA's cBioPortal (www.cbioportal.org) [, ] (). The expressions of COPB2 and NCOR1 have a much higher median than the KRAS expressions but wider standard deviation. This adds to our finding that they are potential KRAS synthetic lethal partners; their high expressions may sustain oncogenic KRAS signaling and knocking them down can have a big effect in cancer cell lines’ viability. On the other hand, SPRY1 has in average lower expressions than KRAS.We further examined the mutation status of KRAS and its three candidate synthetic lethal partners in published colorectal and pancreatic cancer cell lines [, ]. We are expecting that KRAS is very rarely mutated together with the three genes, which sustains the cell line's viability. The results shown in Table prove that this hypothesis holds. […]

Pipeline specifications

Software tools RankAggreg, cBioPortal
Databases GeneCards TCGA Data Portal
Application Gene expression microarray analysis
Diseases Neoplasms, Colorectal Neoplasms