Abstract 3605: ICGC in the cloud
Bibliographic record
Abstract
Abstract In November 2015 members of this consortium and the International Cancer Genome Consortium (ICGC) jointly announced the availability of more than 1,300 whole cancer genomes in the Amazon Web Services’ elastic compute cloud (EC2). Another 480 whole cancer genomes are available in the Cancer Genome Collaboratory, an academic cloud being built by this consortium. By making the data available in cloud compute form, researchers benefit from the high availability, scalability and economy offered by cloud services, and to avoid the large investment in compute resources and the time needed to download the data. Over the next year, we will increase the number of ICGC genomes available in the cloud, with the goal of placing the entire ICGC data set of ∼25,000 donors in academic and commercial clouds when the project is completed in 2018. For information and a getting-started guide, see https://dcc.icgc.org/icgc-in-the-cloud. Cloud computing represents a fundamental shift in the way that cancer genomics is performed. Because of the large size of the ICGC data set, it can take many months to download the data across a typical university broadband connection, and it requires a substantial investment in hardware in order to analyze it. In practice, this has meant that only large computational groups could perform whole-genome analysis at scale. Using the cloud, research groups of any size can launch large analytic processes, pay only for the compute that they use, and avoid charges for data transfer and long-term data storage. A practical demonstration of the power of working in compute clouds comes from our ongoing collaboration with the PanCancer Analysis of Whole Genomes Project (PCAWG; https://dcc.icgc.org/pcawg), which seeks to interpret patterns of variation in both coding and non-coding portions of cancer genomes. Upwards of 2,800 ICGC whole cancer genomes were subjected to a uniform data processing pipeline that included whole genome alignment, uniform quality control, and standardized germline and somatic variant calling using a large number of software packages that were adapted to run efficiently in the cloud. Using a series of 14 academic and commercial compute clouds, we were able to process this 800 terabyte data set in just over a year's time. Given the improvements in the software that occurred over this period, the whole project would take less than 4 months on just a single commercial cloud if we were to start over. When the project is completed later in 2016, we will again use academic and compute clouds to publish the PCAWG data, its major results, and all the software used during the analysis, thereby allowing the research community to integrate PCAWG with their own data sets, and apply the same analytic procedures. Citation Format: Christina K. Yung, Guillaume Bourque, Paul C. Boutros, Khaled El Emam, Vincent Ferretti, Bartha M. Knoppers, Brian O’Connor, B.F. Francis Ouellette, Cenk Sahinalp, Sohrab P. Shah, Lincoln D. Stein, Cancer Genome Collaboratory Consortium. ICGC in the cloud. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 3605.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.003 |
| Science and technology studies | 0.003 | 0.001 |
| Scholarly communication | 0.006 | 0.005 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.002 | 0.004 |
| Insufficient payload (model declined to judge) | 0.130 | 0.095 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".