Parallel genome-scale loss of function screens in 216 cancer cell lines for the identification of context-specific genetic dependencies.
Cowley GS, Weir BA, Vazquez F, Tamayo P, Scott JA, Rusin S, East-Seletsky A, Ali LD, Gerath WF, Pantel SE, Lizotte PH, Jiang G, Hsiao J, Tsherniak A, Dwinell E, Aoyama S, Okamoto M, Harrington W, Gelfand E, Green TM, Tomko MJ, Gopal S, Wong TC, Li H, Howell S, Stransky N, Liefeld T, Jang D, Bistline J, Hill Meyers B, Armstrong SA, Anderson KC, Stegmaier K, Reich M, Pellman D, Boehm JS, Mesirov JP, Golub TR, Root DE, Hahn WC. · 2014
Read Parallel genome-scale loss of function screens in 216 cancer cell lines for the identification of context-specific genetic dependencies. by Cowley GS, Weir BA, Vazquez F, Tamayo P, Scott JA, Rusin S, East-Seletsky A, Ali LD, Gerath WF, Pantel SE, Lizotte PH, Jiang G, Hsiao J, Tsherniak A, Dwinell E, Aoyama S, Okamoto M, Harrington W, Gelfand E, Green TM, Tomko MJ, Gopal S, Wong TC, Li H, Howell S, Stransky N, Liefeld T, Jang D, Bistline J, Hill Meyers B, Armstrong SA, Anderson KC, Stegmaier K, Reich M, Pellman D, Boehm JS, Mesirov JP, Golub TR, Root DE, Hahn WC. (2014) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc4432652.txt.
Introduction
Glenn Cowley and a large team of co-authors shared results of gene shutdown screens in more than two hundred cancer cell lines in Scientific Data in 2014. To grow, each cancer depends on some genes more than others, and those weak points can guide drug design. The team used viruses to deliver short RNA pieces that silence one gene at a time, across the genome.
Each line was tested for how silencing each of thousands of genes changed cell growth after many doublings, read out by sequencing. All lines were screened under the same conditions, so their weak points can be compared fairly. Paired with each line's genetic profile, the data let researchers link a dependency to a mutation or cell type. A software tool provided with the data helps find such links, in straight or curved patterns.
Catalog record
| CREATOR | Cowley GS, Weir BA, Vazquez F, Tamayo P, Scott JA, Rusin S, East-Seletsky A, Ali LD, Gerath WF, Pantel SE, Lizotte PH, Jiang G, Hsiao J, Tsherniak A, Dwinell E, Aoyama S, Okamoto M, Harrington W, Gelfand E, Green TM, Tomko MJ, Gopal S, Wong TC, Li H, Howell S, Stransky N, Liefeld T, Jang D, Bistline J, Hill Meyers B, Armstrong SA, Anderson KC, Stegmaier K, Reich M, Pellman D, Boehm JS, Mesirov JP, Golub TR, Root DE, Hahn WC. |
|---|---|
| YEAR | 2014 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Parallel genome-scale loss of function screens in 216 cancer cell lines for the identification of context-specific genetic dependencies By Glenn S Cowley, Barbara A Weir, Francisca Vazquez, Pablo Tamayo, Justine A Scott, Scott Rusin, Alexandra East-Seletsky, Levi D Ali, William FJ Gerath, Sarah E Pantel, Patrick H Lizotte, Guozhi Jiang Published in: Scientific Data DOI: 10.1038/sdata.2014.35 ABSTRACT Abstract Using a genome-scale, lentivirally delivered shRNA library, we performed massively parallel pooled shRNA screens in 216 cancer cell lines to identify genes that are required for cell proliferation and/or viability. Cell line dependencies on 11,000 genes were interrogated by 5 shRNAs per gene. The proliferation effect of each shRNA in each cell line was assessed by transducing a population of 11M cells with one shRNA-virus per cell and determining the relative enrichment or depletion of each of the 54,000 shRNAs after 16 population doublings using Next Generation Sequencing. All the cell lines were screened using standardized conditions to best assess differential genetic dependencies across cell lines. When combined with genomic characterization of these cell lines, this dataset facilitates the linkage of genetic dependencies with specific cellular contexts (e.g., gene mutations or cell lineage). To enable such comparisons, we developed and provided a bioinformatics tool to identify linear and nonlinear correlations between these features. BACKGROUND & SUMMARY Genome characterization efforts describe an increasingly comprehensive list of genetic alterations that occur in human cancers, but their contributions to the proliferation or survival of cancers often remains obscure. Furthermore, we lack a systematic understanding of the genetic vulnerabilities of human cancer cells as a function of these genetic alterations and the context in which they occur. A complementary effort to systematically assess the genetic requirements of cancer cells in many cellular contexts will help to decipher the roles of specific mutations and the vulnerabilities that these genetic alterations induce. The identification of these context-specific cancer-cell vulnerabilities is the motivation for the generation of the data set described here. To identify genes that have a context-specific effect on cell viability, we performed short hairpin RNA (shRNA) screens in a large number of cell lines in a highly parallel manner, in an effort named Project Achilles. Although screens to identify proliferation/survival genes have been performed in individual cells, Project Achilles data sets are unique in the number of genes and cell lines screened. Genome-wide parallel pooled screens has been performed with 12 (ref. 1), 70 (ref. 2) and 102 (ref. 3) cell lines of a variety of lineages. The data set described here, Achilles 2.4, extends and advances these previously released data sets by increasing the number of cell lines that have been screened and employing a different deconvolution method using next generation sequencing that yields a wider dynamic range and more quantitative assessment than earlier-generation deconvolution methods. The screening pipeline used to create this data set has been previously described3. Briefly, a pool of 54,020 shRNA plasmids targeting ~11,000 genes was used to infect the cell lines with a minimum representation of 200 cells per shRNA in each of 4 replicates. Cells were then propagated for 16 population doublings or 40 days in culture, whichever came first. The relative levels of the shRNA plasmids represented in the cell-pool gDNA were measured using Illumina-based sequencing and compared to the initial plasmid pool (Figure 1). [Figure: Schematic representation of the schema used for pooled shRNA screening.] The Project Achilles data set facilitates the discovery of context-specific dependencies, genes that when suppressed only have an effect on the viability of the particular cell lines, combined with the discovery of biomarkers associated with these differential sensitivities. Using Project Achilles 2.0 (ref. 3) and Project Achilles 2.4, the ‘oncogene addictions’ of many known oncogenes such as PIK3CA, KRAS, and BRAF are identified by showing that cell lines harboring mutations of these genes exhibit higher sensitivity to their suppression. Furthermore, presence of these oncogenes in cells predicts dependency on other genes, for example PIK3CA mut cell lines have preferential dependency on MTOR. In addition to these known dependencies, we recently identified SMARCA2 and ARID1B as essential genes in cell lines with SMARCA4 and ARID1A mutations, respectively4,5. In addition to these single-gene based relationships, the Achilles data set can be exploited to discover pathways to which specific tumor subtypes are preferentially dependent. One example of using such approach is the discovery of the dependency of WNT active tumors on members of the YAP pathway6. By making the data and analytical tools available to the scientific community, we expect that a number of new vulnerabilities will be revealed. We made Achilles 2.4 available to the scientific community in the Project Achilles portal (www.broadinstitute.org/achilles). In addition to the shRNA level data, a gene-level dataset generated using the ATARiS algorithm7 is also available for download. ATARiS combines data from the multiple distinct shRNAs per target gene across many cell lines to reduce the contribution of off-target effects while reinforcing the on-target effects. Moreover, specific genes can be queried and the dependency profile for such genes across cell lines can be individually downloaded from the portal. To facilitate the discovery of molecular and cellular correlates of the dependencies profiles, we created an analytical tool that we named PARIS (Probability Analysis by Ranked Information Score) and made it available to the scientific community in GenePattern (www.genepattern.org). METHODS Cell line information The majority of cell lines (179) were obtained from the Cancer Cell Line Encyclopedia (www.broadinstitute.org/ccle). Cell line information, including source is listed in ’Table1_screening_information.xls’ (Data Citation 1). Tumor type and growth media conditions, also used for screening, were obtained from the CCLE project. For cell lines not obtained from CCLE, media conditions used by the source laboratory were employed. Cell doubling time is calculated from the lentivirally infected cells during the course of the experiment. Days in culture, calculated from the date of infection until the date of the harvest, and passage number, based on the number of splits that occurred during the time in culture, refer to the time point of the sample that was used for data collection specific to each cell line. SNP fingerprinting To ensure the identity of the cell lines and rule out the possibility of cross-contamination during the screening process, a SNP fingerprinting quality control step was implemented. SNP fingerprinting matches a panel of reference SNP genotypes for a cell line, with genotypes assayed after the screening process. The reference set of SNP genotypes used for most samples was derived from the Affymetrix SNP6.0 array birdseed genotypes from the Cancer Cell Line Encyclopedia project8,9. Cell lines not present in CCLE were genotyped for the same panel of SNPs with either the Sequenom or Fluidigm platform and/or also profiled with SNP6.0 arrays before screening. Two genotyping platforms, Sequenom and Fluidigm, were used to assay different panels of SNPs, which overlapped those present on the Affymetrix SNP6.0 array. DNA was isolated as described in the screening methods (below). Briefly the Sequenom protocol is as follows: SNPs are amplified in PCR reactions that contain a maximum of 24 loci. The Single Base Extension reaction is then performed on the Shrimp Alkaline Phosphatase treated PCR product using iPLEX-GOLD enzyme and mass-modified terminators (Sequenom, SanDiego). A small volume of reaction is then loaded onto each position of