GSCAN--or the GWAS & Sequencing Consortium of Alcohol and Nicotine use--is an international genetic association meta-analysis consortium. Our goal is to aggregate genetic association findings across scores of studies with millions of individuals. GSCAN is composed of three independent but related projects: 1) an exome chip meta-analysis of low-frequency non-synonymous variants, 2) a GWAS meta-analysis, and 3) a whole genome sequencing association meta-analysis.
- 1 Summary Statistics
- 2 Meetings
- 3 GSCAN Exome Chip
- 4 GSCAN GWAS Phase 1
- 5 GSCAN GWAS Phase 2
- 6 GSCAN Sequencing
- 7 Authorship guidelines
Summary statistics from Liu et al. (2019) "Association Studies of up to 1.2 million individuals yield new insights into the genetic etiology of tobacco and alcohol use" published in Nature Genetics. are archived at the University of Minnesota library:
Please contact the dataset contact on that page if you notice any issues. We will readily fix them.
Regular conference calls are held and minutes are available here.
Other meeting materials from internal meetings are here:
GSCAN Exome Chip
Phenotype definitions and analysis plan
Freeze 1. We concluded a pilot freeze of the exome chip project in 2015 and are writing up our results now. All of the summary statistics are on twins at /net/twins/svrieze/everything-else/wp/GSCAN/freeze1-25-Mar-2015.
Freeze 2. New studies that will be included in Freeze 2 are located on RC at /work/KellerLab/GSCAN/EXOME. Each folder in that directory is the name of a study and includes two subfolders, one for Phenotypes and one for Genotypes. Genotypes are split by chromosome to facilitate analyses.
GSCAN GWAS Phase 1
- Phenotype workgroup: Laura Bierut, Marilyn Cornelis, Dave Hinds, Youna Hu, Jaakko Kaprio, Eric Jorgenson, Dajiang Liu, Matt McGue, Marcus Munafo, Gunter Schumann, Scott Vrieze, Luisa Zuccolo
- Analysis workgroup: Goncalo Abecasis, David Hinds, Youna Hu, Eric Jorgenson, Charles Kooperberg, Pete Kraft, Penelope Lind, Dajiang Liu, Nancy Saccone, Dan Stram, Scott Vrieze, Xiaowei Zhan
Phenotype definitions and analysis plan
The analysis plan and phenotypes are described in files linked below (makes it easier to keep track of versioning!). Coding of phenotypes is described in the aptly-named "phenotype definitions" file whereas the genome-wide analysis plan is in the all-too-aptly-named "analysis plan" document. Please note that the phenotype definitions document only contains information on how to code the eight smoking/drinking phenotypes. File formats for those phenotypes, which many will recognize as standard pedigree formats, are included in the analysis plan. Everything else should be fairly straightforward.
Coordination and organization
Study contact info is tracked in this Google Sheet.
Studies available in dbGaP, along with accession numbers, etc. are tracked in this Airtable.
Study data to which we have direct access are located either on twins or RC. Twins data are organized in the folder /net/twins/svrieze/everything-else/wp/GSCAN/GWAS. Within this folder those studies to which we have raw data access are in the folder CU_Boulder_samples (for lack of a better name!). Summary stats generated on these samples are organized within summary_stats_generated_internally. Summary stats generated by outside groups and submitted for meta-analysis are organized within summary_stats_generated_externally.
On RC the organization is similar. Everything is located within the folder /work/KellerLab/GSCAN/GWAS. Study data to which we have raw data access are in the folder individual_level_study_data. Summary stats generated on these samples are organized within summary_stats_generated_internally. Summary stats generated by outside groups and submitted for meta-analysis are organized within summary_stats_generated_externally.
GSCAN dbGaP & UK Biobank
Studies included from dbGaP, and the process by which phenotypes and genotypes were constructed and merged is outlined on the GSCAN dbGaP page.
GSCAN GWAS Phase 2
In April 2018 we began working on a second phase of GSCAN, including many more studies of diverse ancestry.
Our internal document to track progress | is here.
We hope to update this section with detailed descriptions of how we have conducted phenotype derivations for each TOPMed cohort to which we have access to raw data.
- For now, the R scripts to go from source phenotype file to eventual derived phenotype is located here:
- We're tracking analyses in this Google doc
Phenotype definitions and analysis plan for external studies
Phenotype definitions and analysis plans for the TOPMed studies are contained in this document.
The list of dbGaP studies in TOPMed is in this Airtable.
While authorship is decided on an individual basis for each GSCAN paper, typically, authorship is arranged in groups. We hope the GIANT investigators will forgive us for adopting their authorship guidelines.
- A group of 6 or fewer junior investigators who strongly led the efforts, usually starred to denote equal contribution, followed by additional junior investigators who played key, central roles.
- In alphabetical order, junior investigators who had substantial individual contributions but not as much as those in Group 1. Typically, these might be lead analysts or other junior investigators who made a sizable contribution such as GWA analyses performed specifically for the paper.
- In alphabetical order, junior investigators who had notable individual contributions but not as much as those in Groups 1 or 2. Typically, these might be lead analysts for replication cohorts, providing results for a group of top hits.
- In alphabetical order, junior and senior investigators who had contributions worthy of authorship (participating in analysis, phenotype collection, genotyping, oversight of cohorts, etc. that was specific to the paper) but not as much as those in the other groups.
- In alphabetical order, senior investigators who had contributions worthy of authorship and contributed more than those in group 4. Typically, these might be a lead PI of a participating cohort who did not participate as strongly in GSCAN activities as those in group 6.
- In alphabetical order, senior investigators who participated strongly in GSCAN activities but did not strongly lead/oversee the writing and/or analysis for the paper. Typically, these might be leaders of key GSCAN activities.
- The senior investigators who strongly led/oversaw the writing and/or analysis of the paper, including a subset that are co-corresponding authors (usually 6 or fewer).
Proposal for number of authors
Given the author categories above, we expect each study to have 3-7 or fewer "study-specific" authors. Study-specific authors include local analysts who generated summary stats as well as PIs or others who supervised the local study work. These authors are in addition to any individuals associated with those studies who contributed in significant ways to the central meta-analysis or writing. We will use a tiered number of study-specific authors that tracks along sample size, given that larger studies often require more people to contribute to data generation/analysis.
Note: The Ns below are total sample size across all phenotypes reported in a given paper. (A study that has 10,000 samples for cigs per day and 20,000 for smoking initiation would have N=30,000.)
- 3 authors for N < 10,000
- 4 authors for N of 10,000-50,000
- 5 authors for N of 50,000-100,000
- 6 authors for N of 100,000+