Personalized Environment and Genes Study (PEGS) Dataset-a resource for genomic, exposomic, and geospatial data.

PEGS comprises a compatible, multi-dimensional collection of datasets in consistent and programmatically extractable formats, as shown in Fig.  2 . Survey-based data on exposures and medication, geospatial data, and genomic data are available for PEGS participants. Table  5 outlines the data components of PEGS, including sample sizes for PEGS Data Freeze 3.1. Thorough quality control was conducted to ensure consistent formatting, nomenclature, and data encoding. Metadata were added to each PEGS data component to specify the data format, data encoding, special codes, descriptions, and structure of survey questions to enable the identification of nested questions. There is extensive documentation for each PEGS data component describing the methods used to create the data and details of the variables. PEGS data are available for analysis in R (.RData), SAS (.sas7bdat), and text (.csv) file formats. The quality control performed for the data and its availability in multiple formats facilitates consistent, reproducible, and comparable analyses and supports research collaborations. As of June 27, 2023, PEGS Data Freeze 3.1 is the most recent version released. PEGS data are updated quarterly with data from additional participants, updated data from existing participants, data on new variables, and additional data components.

PEGS data can empower research in a broad range of areas. The data are readily available to confirm or replicate discovery results. Investigators can conduct ExWAS in PEGS data to uncover novel associations between endogenous and exogenous exposures and disease risk, in both single- and multi-exposure analyses. In addition, the data can be used to develop polyexposure risk scores and compare their performance to that of traditional polygenic and clinical risk scores for disease risk prediction 21 . The data can also be used to conduct variance decomposition of disease risk to assess the proportion of disease risk that genetics (G), the environment (E), SDOH (S), and their interactions (GxE, GxS, ExS, GxExS) explain. Additionally, PEGS data can be utilized to explore genetic correlations with environmental exposures, including common variants, rare variants, gene-based associations, and pathway-based analyses, to unravel these complex relationships. Further, incorporating geospatial data enables research on how genomics, the environment, and SDOH contribute to disease risk. Geospatial estimates of exposure, including air quality and proximity to pollutant sources, enable the examination of how weather and climate exposures such as extreme heat are related to disease risk and progression. Given the diversity and dimensionality of PEGS data, this resource can serve as a testing ground for innovative statistical methods. For example, PEGS data was used in a Sage Bionetworks DREAM Challenge ( https://dreamchallenges.org/ ). DREAM Challenges are collaborative, competitive scientific initiatives that bring together researchers, data scientists, and experts from various fields to address complex biomedical and healthcare problems. Participants are tasked with developing innovative computational models and solutions to solve specific challenges in areas such as disease classification. NIEHS continues to expand its investment in PEGS through ongoing efforts. Work began in 2019 to link PEGS data with electronic health records (EHRs) of patients of the Duke University Health System and UNC Health at the University of North Carolina Chapel Hill. Linking EHRs for consenting participants will enable the integration of temporal health and disease data and clinical information on multi-dimensional phenotypes, such as International Classification of Diseases (ICD) codes, laboratory data, images, and vital signs. A large call-back study is in progress to collect a range of biospecimens, including blood, urine, stool, saliva, nasal cells, hair, nail clippings, baby teeth, and household dust. A major goal of PEGS is supporting collaborative research by sharing data with researchers engaged in collaborative projects. Information about submitting proposals for collaborative research is available at: https://www.niehs.nih.gov/research/atniehs/labs/crb/studies/pegs/collaboration/proposal . Several available tools enable users to explore PEGS data and results from analyses conducted in the data. The Informatics for Integrating Biology and the Bedside (i2b2) Query and Analysis Tool facilitates the design and feasibility assessment of potential analyses by allowing researchers to explore aggregated PEGS data 28 . The i2b2 Query and Analysis Tool provides approved users with easy access to explore basic statistics from de-identified and aggregated PEGS data by building customizable queries to display tabular and graphical summaries. The PEGS Explorer web application ( https://pegsexplorer.niehs.nih.gov/ ) was developed to share published results of ExWAS conducted in PEGS data and enable the exploration of rigorously calculated exposure correlations 29 , 30 . Through globe visualizations, PEGS Explorer users can explore and visualize correlations between exposures associated with common, complex diseases. Ongoing work in data analysis is focused on building analysis pipelines and workflows to enable efficient, insightful, and collaborative research and ensure consistent, reproducible, and comparable analyses.

PEGS participants complete three surveys to provide phenotype and exposure data. First administered in 2013, the PEGS Health and Exposure Survey (N = 9,449) collects data on general demographics, family medical history, lifestyle factors such as smoking and alcohol use, and occupational exposures. Beginning in 2017, the NIEHS PEGS Exposome Survey was also administered to collect comprehensive information about endogenous and exogenous exposures throughout life. The External Exposome Survey (Exposome Part A) (N = 3,618) is focused on external exposures, including chemical and environmental exposures at work and home from childhood to the present. The Internal Exposome Survey (Exposome Part B) (N = 3,071) is focused on internal exposures, including medications and lifestyle factors such as physical activity, stress, sleep, and diet. Table  2 outlines the question categories in the Health and Exposure Survey and the Internal and External Exposome Surveys, and Figs.  4 – 6 provide snapshots of the types of information requested by each survey. Additional screener surveys for diabetes and eczema were administered to 227 and 329 participants, respectively. The surveys are available for download on the PEGS website ( https://www.niehs.nih.gov/research/atniehs/labs/crb/studies/pegs/about/data ). Study protocol and other details can be found on Clinicaltrials.gov (ClinicalTrials.gov ID: NCT00341237 ). Table 2 Categories of questions in the PEGS Health and Exposure Survey, External Exposome Survey, and Internal Exposome Survey. Health and Exposure Survey About Your Family’s Health Diabetes and Endocrine Neurologic About Your General Health Digestive Occupation About Your Home Life Exposures Renal About Your Mood Fatigue Reproductive (Females Only) Bones, Joints, and Muscles Hematological Reproductive (Males Only) Cancer Immune Respiratory Cardiovascular Lifestyle Skin, Eyes, and Hair External Exposome (Exposome A) Characteristics of Current and Past Residences: • Agricultural Property Use • Garage and Basement • Heating and Cooling • Pesticides and Insecticides • Pets • Surrounding Area • Walls and Flooring • Water and Dampness Chemical and Metal Exposures at Work Ultraviolet Light Exposure Workplace Characteristics Hobby Exposures Internal Exposome (Exposome B) Chemotherapy/Radiation Therapy Physical Activity Dietary Behavior Reproductive History (Females Only) Dietary Intake Sleep Genetic History Stress Infectious Disease Vitamins, Minerals, and Other Supplement Use Medications Twin/Triplet Siblings and Birth Order Other High-level survey question categories administered in the PEGS Health and Exposure Survey, External Exposome Survey, and Internal Exposome Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Fig. 4 Percentages of participants with selected self-reported diseases or conditions and lifestyle factors from the PEGS Health and Exposure Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Fig. 5 Percentages of participants exposed to selected external environmental exposures from the PEGS External Exposome Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Fig. 6 Percentages of PEGS participants exposed to selected internal environmental exposures from the PEGS Internal Exposome Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Categories of questions in the PEGS Health and Exposure Survey, External Exposome Survey, and Internal Exposome Survey. Characteristics of Current and Past Residences: • Agricultural Property Use • Garage and Basement • Heating and Cooling • Pesticides and Insecticides • Pets • Surrounding Area • Walls and Flooring • Water and Dampness High-level survey question categories administered in the PEGS Health and Exposure Survey, External Exposome Survey, and Internal Exposome Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Percentages of participants with selected self-reported diseases or conditions and lifestyle factors from the PEGS Health and Exposure Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Percentages of participants exposed to selected external environmental exposures from the PEGS External Exposome Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Percentages of PEGS participants exposed to selected internal environmental exposures from the PEGS Internal Exposome Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Questionnaire data underwent structured quality control procedures, including range validation, logical consistency checks across related questions, verification of programmed skip patterns, and harmonization of categorical encodings across survey waves. Implausible responses were flagged based on predefined thresholds and reviewed. Free-text responses were standardized using rule-based parsing followed by manual review when necessary. Comprehensive data dictionaries and coding schemas accompany each PEGS Data Freeze to facilitate reproducible analysis. Self-reported data collected with the Health and Exposure Survey and the Exposome Surveys have been aggregated for case counts for numerous phenotypes. Tables  3 and 4 list self-reported phenotypes with a minimum of 100 cases among all participants and among participants for whom WGS data are available. Table 3 Number of cases for self-reported diseases and conditions in PEGS participants. Disease diagnosis Cases-all participants Disease diagnosis Cases-all participants Ever diagnosed with allergic rhinitis, hay fever, or seasonal allergies 3751 Cancer diagnosis - skin (non-melanoma) 508 Ever diagnosed with high blood pressure or hypertension 3460 Ever diagnosed with stomach or duodenal ulcer 502 Ever diagnosed with high cholesterol 3094 Diagnosed with manic-depressive illness or bipolar disorder 488 Ever diagnosed with allergies or allergic reactions (other than seasonal allergies) 2620 Ever diagnosed with rheumatoid arthritis (RA) 470 Ever diagnosed with chicken pox 2598 Ever diagnosed with chronic obstructive pulmonary disease (COPD) (e.g., chronic bronchitis, emphysema) 467 Ever diagnosed with flu 2062 Ever diagnosed with a sleep disorder 465 Ever diagnosed with migraine headaches (with or without aura) 1649 Males only - ever diagnosed with enlarged prostate or benign prostatic hyperplasia 460 Ever diagnosed with iron-deficiency anemia 1524 Ever diagnosed with viral food poisoning 458 Ever diagnosed with polyps in the colon or rectum 1469 Ever diagnosed with pyelonephritis, nephritis, or kidney infection 441 Females only - ever diagnosed with fibroids, fibroid tumors, uterine fibroids, or other benign uterine tumors 1457 Ever diagnosed with post-traumatic stress disorder (PTSD) 434 Any cancer diagnosis 1321 Ever diagnosed with a staph infection 432 Ever diagnosed with osteoarthritis 1247 Ever diagnosed with brittle bones or osteoporosis 426 Ever diagnosed with bone loss, thinning of the bones, osteopenia, or pre-osteoporosis 1240 Cancer diagnosis - had surgery 401 Ever diagnosed with asthma 1233 Ever diagnosed with psoriasis 399 Females only - ever diagnosed with ovarian cysts or benign ovarian growth or neoplasm 1224 Ever diagnosed with fibromyalgia 383 Ever diagnosed with diabetes or sugar diabetes 1142 Ever diagnosed with gout 364 Ever diagnosed with pre-diabetes, impaired fasting glucose, or impaired glucose tolerance 1126 Sleep disorder diagnosis - sleep apnea 362 Ever diagnosed with measles 1109 Females only - ever diagnosed with polyps in the endometrium or uterus 351 Ever diagnosed with thyroid disease (other than cancer) 1097 Ever diagnosed with shingles 350 Ever diagnosed with mumps 1000 Ever diagnosed with poor blood flow or blocked or narrowed arteries to the legs, claudication, or peripheral arterial disease 346 Ever diagnosed with cold sores 954 Ever diagnosed with diabetes - currently being treated with insulin 337 Ever diagnosed with another type of arthritis 912 Ever diagnosed with coronary artery disease 312 Ever diagnosed with eczema 877 Ever diagnosed with streptococcal invasive disease 297 Ever diagnosed with kidney stones 874 Ever diagnosed with heart attack or myocardial infarction (MI) 290 Ever diagnosed with shingles 834 Ever diagnosed with genital warts 275 Ever diagnosed with viral pneumonia 781 Cancer diagnosis - breast, including ductal carcinoma in situ (DCIS) 258 Ever diagnosed with German measles 773 Ever diagnosed with hepatitis 255 Ever diagnosed with gallbladder disease 752 Ever diagnosed with chlamydia 251 Ever diagnosed with urticaria or hives 734 Males only - ever diagnosed with inflammation of the prostate or prostatitis 242 Ever diagnosed with diabetes - currently using diabetic pills (oral medication) to lower blood sugar 729 Ever diagnosed with hyperthyroidism (e.g., Grave’s disease) 239 Ever diagnosed with lactose intolerance 712 Diagnosed with chronic fatigue syndrome by a doctor or health care professional 229 Ever diagnosed with hypothyroidism (e.g., Hashimoto’s thyroiditis) 703 Ever diagnosed with ulcers 220 Females only - ever diagnosed with endometriosis 661 Ever diagnosed with fatty liver disease or steatosis 214 Ever diagnosed with bacterial food poisoning 605 Ever diagnosed with Raynaud’s syndrome or disease 211 Ever diagnosed with cardiac arrhythmia 561 Ever diagnosed with enlarged thyroid or goiter 205 Ever diagnosed with bacterial pneumonia 534 Ever diagnosed with genital herpes 196 Ever diagnosed with mononucleosis (mono) 522 Ever diagnosed with whooping cough 196 Includes all diagnoses with at least 100 cases among all PEGS participants, arranged in decreasing order of number of cases among all participants. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Table 4 Number of cases for self-reported diseases and conditions in PEGS participants. Disease diagnosis Cases-participants with whole-genome sequencing Disease diagnosis Cases-participants with whole-genome sequencing Ever diagnosed with allergic rhinitis, hay fever, or seasonal allergies 2050 Cancer diagnosis - skin (non-melanoma) 303 Ever diagnosed with high blood pressure or hypertension 1614 Ever diagnosed with stomach or duodenal ulcer 259 Ever diagnosed with high cholesterol 1617 Diagnosed with manic-depressive illness or bipolar disorder 188 Ever diagnosed with allergies or allergic reactions (other than seasonal allergies) 1427 Ever diagnosed with rheumatoid arthritis (RA) 193 Ever diagnosed with chicken pox 2070 Ever diagnosed with chronic obstructive pulmonary disease (COPD) (e.g., chronic bronchitis, emphysema) 189 Ever diagnosed with flu 1629 Ever diagnosed with a sleep disorder 374 Ever diagnosed with migraine headaches (with or without aura) 880 Males only - ever diagnosed with enlarged prostate or benign prostatic hyperplasia 243 Ever diagnosed with iron-deficiency anemia 827 Ever diagnosed with viral food poisoning 371 Ever diagnosed with polyps in the colon or rectum 816 Ever diagnosed with pyelonephritis, nephritis, or kidney infection 227 Females only - ever diagnosed with fibroids, fibroid tumors, uterine fibroids, or other benign uterine tumors 746 Ever diagnosed with post-traumatic stress disorder (PTSD) 208 Any cancer diagnosis 693 Ever diagnosed with a staph infection 341 Ever diagnosed with osteoarthritis 707 Ever diagnosed with brittle bones or osteoporosis 170 Ever diagnosed with bone loss, thinning of the bones, osteopenia, or pre-osteoporosis 666 Cancer diagnosis - had surgery 338 Ever diagnosed with asthma 607 Ever diagnosed with psoriasis 208 Females only - ever diagnosed with ovarian cysts or benign ovarian growth or neoplasm 660 Ever diagnosed with fibromyalgia 171 Ever diagnosed with diabetes or sugar diabetes 481 Ever diagnosed with gout 155 Ever diagnosed with pre-diabetes, impaired fasting glucose, or impaired glucose tolerance 567 Sleep disorder diagnosis - sleep apnea 291 Ever diagnosed with measles 920 Females only - ever diagnosed with polyps in the endometrium or uterus 192 Ever diagnosed with thyroid disease (other than cancer) 595 Ever diagnosed with shingles 282 Ever diagnosed with mumps 842 Ever diagnosed with poor blood flow or blocked or narrowed arteries to the legs, claudication, or peripheral arterial disease 136 Ever diagnosed with cold sores 771 Ever diagnosed with diabetes - currently being treated with insulin 118 Ever diagnosed with another type of arthritis 385 Ever diagnosed with coronary artery disease 141 Ever diagnosed with eczema 468 Ever diagnosed with streptococcal invasive disease 232 Ever diagnosed with kidney stones 429 Ever diagnosed with heart attack or myocardial infarction (MI) 112 Ever diagnosed with shingles 420 Ever diagnosed with genital warts 216 Ever diagnosed with viral pneumonia 606 Cancer diagnosis - breast, including ductal carcinoma in situ (DCIS) 139 Ever diagnosed with German measles 659 Ever diagnosed with hepatitis 129 Ever diagnosed with gallbladder disease 390 Ever diagnosed with chlamydia 184 Ever diagnosed with urticaria or hives 436 Males only - ever diagnosed with inflammation of the prostate or prostatitis 132 Ever diagnosed with diabetes - currently using diabetic pills (oral medication) to lower blood sugar 292 Ever diagnosed with hyperthyroidism (e.g., Grave’s disease) 119 Ever diagnosed with lactose intolerance 363 Diagnosed with chronic fatigue syndrome by a doctor or health care professional 100 Ever diagnosed with hypothyroidism (e.g., Hashimoto’s thyroiditis) 420 Ever diagnosed with ulcers 188 Females only - ever diagnosed with endometriosis 351 Ever diagnosed with fatty liver disease or steatosis 107 Ever diagnosed with bacterial food poisoning 485 Ever diagnosed with Raynaud’s syndrome or disease 134 Ever diagnosed with cardiac arrhythmia 308 Ever diagnosed with enlarged thyroid or goiter 100 Ever diagnosed with bacterial pneumonia 418 Ever diagnosed with genital herpes 146 Ever diagnosed with mononucleosis (mono) 427 Ever diagnosed with whooping cough 169 Includes all diagnoses with at least 100 cases among PEGS participants with available whole-genome sequencing data, arranged in decreasing order of number of cases among all participants. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Number of cases for self-reported diseases and conditions in PEGS participants. Includes all diagnoses with at least 100 cases among all PEGS participants, arranged in decreasing order of number of cases among all participants. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Number of cases for self-reported diseases and conditions in PEGS participants. Includes all diagnoses with at least 100 cases among PEGS participants with available whole-genome sequencing data, arranged in decreasing order of number of cases among all participants. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Self-reported free-text data on medications collected with the Internal Exposome Survey have been translated into Anatomical Therapeutic Chemical (ATC) codes per the World Health Organization’s ATC classification system ( https://www.who.int/tools/atc-ddd-toolkit/atc-classification ). Briefly, the self-reported free-text medication data were cleaned and then matched to standardized medication names in the Food and Drug Administration’s National Drug Code Directory to find active ingredients for each listed medication name. These active ingredients were then matched to ATC codes from the 2021 ATC/DDD Index ( https://www.whocc.no/atc_ddd_index/ ). Medications that did not match available standardized medication names were manually cleaned for spelling mistakes and regional differences in medication names or were encoded as supplements when appropriate. Drugs that did not match available ATC codes were included in a broader ATC code category when possible. The most complete level of ATC code and ATC name, if available, were mapped. A limitation of the questionnaire-based medication data is the unavailability of the route of administration and dosing. Table  5 outlines the data available for the PEGS cohort. Table 5 PEGS data components. Category and Components Description Participants Survey Data Demographic and administrative data Demographics, consent, address, and administrative data for all participants 19445 Health and Exposure Survey Demographics, health, family history of disease, environmental exposures, socioeconomic status, and lifestyle 9449 External Exposome Survey (Exposome A) Residential and occupational environmental exposures 3618 Internal Exposome Survey (Exposome B) Medication use, physical activity, stress, sleep, diet, genetics, and reproductive history 3071 Diabetes Screener Survey Diabetes screener administered to participants with self-reported diabetes 227 Eczema Screener Survey Eczema screener administered to participants with self-reported eczema 329 Right-not-to-know Main Survey Right-not-to-know survey administered for incidental findings and reports 231 Right-not-to-know Cognitive Interview Survey Right-not-to-know cognitive interview administered to assess awareness of incidental findings and reports 12 Medication Data Anatomical Therapeutic Chemical (ATC) codes ATC codes for self-reported free-text medication names from the Internal Exposome Survey (Exposome B) per the World Health Organization ATC classification system 2263 Genomic Data Candidate gene/single-nucleotide polymorphism (SNP) data Candidate SNP data for a subset of participants for specific research goals 12316 Single nucleotide variants (SNVs) SNV and small indel genotypes derived from the whole-genome sequencing (WGS) data in PLINK’s.bed/.bim/.fam format 4737 Structural variants Structural variant calls generated from the whole-genome sequencing (WGS) data in.vcf format consisting of large deletions, duplications, and inversions 4737 Human leukocyte antigen (HLA) genotypes HLA genotypes identified from the WGS data for 20 HLA genes with up to six digits of specificity 4737 Telomeric content Aggregate telomeric content estimated from WGS reads reported as telomeric reads per GC content-matched million reads 4737 Local and global ancestry estimations Inferred local ancestry per chromosome after haplotype phasing and global estimates of percent ancestry for each participant 4730 Methylation data Genome-wide methylation profiling data using the Infinium MethylationEPIC v1.0 BeadChip Kit targeting 866,297 CpG sites 4724 Geospatial Data Geocodes (GIS) Geocoded participant addresses from five study events with mapping coordinates 18462 Hazards data Exposure estimates and proximity measures calculated using geospatial linkages from the following databases: Atmospheric Composition Analysis Group (ACAG), Toxics Release Inventory (TRI), Center for Air, Climate, and Energy Solutions (CACES), North Carolina Department of Environmental Quality (NCDEQ), Department of Transportation (DOT), Federal Aviation Administration (FAA), Federal Communications Commission (FCC), and the Nuclear Regulatory Commission (NRC) 18462 MERRA-2 data (Earthdata) Geospatial data linkages from the Modern Era Retrospective Analysis for Research and Applications (MERRA-2) project containing consistent estimates of climate and environmental metrics from a range of satellite-based environmental observations 17273 Social Vulnerability Index (SVI) data Geospatial data linkages for Centers for Disease Control/ Agency for Toxic Substances and Disease Registry (CDC/ATSDR) SVI containing summaries of social determinants of health at the census-tract level 17273 Environmental Justice Index (EJI) data Geospatial data linkage for CDC/ATSDR EJI containing summaries of environmental, social, and health factors at the census-tract level 17273 Data components available for the PEGS cohort. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. PEGS data components. Data components available for the PEGS cohort. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Self-reported income categories reflect nominal income at the time of survey completion and were not adjusted for inflation. Because specific survey dates were removed to protect participant privacy, users cannot retrospectively adjust income variables for inflation; analyses should interpret income categories as contemporaneous to survey completion. Initial genotyping of participant samples was completed as part of clinical call-back studies. In 2010, the pilot EPR Consortium Project genotyped 87 environmental response genes (656 single-nucleotide polymorphisms (SNPs) in approximately 3,700 randomly selected participants). Additionally, as part of genotype collaboration efforts initiated in 2014 to address specific research goals, genotyping was performed for 12,316 participants for targeted candidate genes and SNPs. In 2019, the Broad Institute performed WGS using blood samples from 4,737 PEGS participants with the most complete survey data to interrogate common and rare variants and structural variations, including high-resolution human leukocyte antigen (HLA) variants. Paired-end Illumina short-read sequencing was utilized for this purpose with a target depth of greater than 30x. The WGS data were aligned to the hg38 human reference assembly to obtain single-nucleotide variants and small insertions/deletions (indels). After quality control, approximately 43 million high-quality variants for 4,607 participants were annotated with the WGS Annotator. WGS also identified HLA genotypes with up to six digits of specificity for four- and six-digit HLA alleles for six common HLA gene types (HLA-A, -B, -C, -DQA, DQB, and -DRB) and 14 additional HLA genes (HLA-DOA, -DOB, -DMA, -DMB, -DPA1, -DPB1, -DRQ, -DRB3, -DRB5, -F, -G, -H, -J, and -L). A total of 92,297 structural variants consisting of large deletions, duplications, and inversions, as well as estimated total telomeric content for each participant, were also identified from the WGS data. Figure  8 and Table  5 summarize the genomic data available for the PEGS cohort. Fig. 7 PEGS GIS data. Overview of multi-pollutant point sources used to estimate exposure indicators for PEGS participants using their geocoded addresses. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Fig. 8 Summary of the genomic data available in the PEGS cohort. Whole-genome sequencing data were used to obtain single-nucleotide variants consisting of common and rare variants, structural variant calls, human leukocyte antigen (HLA) genotypes for 20 HLA genes, aggregate telomeric content estimation, inferred local ancestry per chromosome, and global estimates of percent ancestry. Genome-wide methylation profiling data were obtained using the Infinium MethylationEPIC v1.0 BeadChip Kit. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. PEGS GIS data. Overview of multi-pollutant point sources used to estimate exposure indicators for PEGS participants using their geocoded addresses. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Summary of the genomic data available in the PEGS cohort. Whole-genome sequencing data were used to obtain single-nucleotide variants consisting of common and rare variants, structural variant calls, human leukocyte antigen (HLA) genotypes for 20 HLA genes, aggregate telomeric content estimation, inferred local ancestry per chromosome, and global estimates of percent ancestry. Genome-wide methylation profiling data were obtained using the Infinium MethylationEPIC v1.0 BeadChip Kit. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. The Broad Institute sequenced all 4,737 samples on the NovaSeq6000 platform with a target genome-wide read depth of 30× (24 samples per flow cell). The Broad Institute used their gatk4-genome-processing-pipeline version 1.0.0 (github.com/gatk-workflows/gatk4-genome-processing-pipeline) to process reads 7 . Using bwa mem 0.7.15-r1140 8 , the reads were aligned to the hg38 reference genome. Using Picard Tool MarkDuplicates 2.20.4, duplicate fragments were flagged. Finally, using GATK 4.0.10.1 ApplyBQSR, base quality score recalibration was performed. Variant calling was performed per sample using GATK 3.5.0 HaplotypeCaller with output in g.vcf format, and joint genotyping was performed using GATK 4.1.4.0 GenotypeGvcfs. Subsequently, GATK 4.1.1.0 VariantFiltration was used to apply filtering flags, and GATK 4.1.1.0 ApplyVQSR was utilized for variant quality score recalibration. A “PASS” filter label was assigned to variants with an excess heterozygosity phred score of <  = 54.69 and a VQSLOD score exceeding the threshold of retaining 99.0% of true positive indels and 99.7% of true positive SNPs. Samtools flagstats 9 was used to determine the total read count and proportion of duplicate fragments. samtools idxstats was used to determine the percentages of reads mapping to the mitochondrial genome and chrY, as well as the ratio of reads mapped to chrX per bp, relative to the genome-wide rate. FastQC 0.11.9 was used to determine the aggregate %GC, separately for mate 1 and mate 2 reads. From the FastQC output, the maximum deviation of mean per-cycle sequence content from genomic background rates of 20% for G/C and 30% for A/T was also determined, excluding the first nine cycles due to the anticipation of large deviations. Per-sample percentages of reads with mapping quality below 5 and 10 were directly derived from the MAPQ field of sample CRAM files. Counts of SNPs and indel variants were derived from preliminary unfiltered GATK joint genotypes called in batches of samples sequenced and processed in tandem. From these preliminary calls, the mean and standard deviation of variant genotype qualities, as well as major allele frequencies, were determined. The major allele frequencies were calculated separately for homozygous and heterozygous genotypes. Estimates of genomic proportions derived from Asian, Amerindian, European, and African ancestors were determined using ancestry informative markers described by Kosoy et al . 10 and the genotypes of individuals with known ancestry reported in Table S6 of that publication. Genotypes called at these loci for each sample were extracted and independently merged with individuals of known ancestry using PLINK 1.90b4.3 11 and then analyzed using fastStructure version 1.0 12 with four specified ancestral populations (-K 4). fastStructure produces genome proportion estimates derived from unlabeled ancestral populations inferred from the given data. Continental associations for each estimated population were determined by calculating the mean proportions separately for individuals of known ancestry and pairing the proportions with a value of 0.9 or higher. In rare cases for which fastStructure’s non-deterministic algorithm did not separate individuals with known ancestry such that each known population could be assigned as described to a single inferred population, the analysis was repeated. Some individual samples had higher than expected GC content and fractions of reads mapped to chrY. In addition, a small proportion of the genome exceeded the target of 30× coverage. However, these features did not appear to be associated with an increase in novel variant calls or a reduction in genotype quality estimates and hence did not require the removal of any samples. Estimated ancestral genome proportions were manually compared with self-reported race and ethnicity, and ratios of chrX to genomic coverage and fractions of reads mapped to chrY were examined in the context of self-reported sex. These observations indicated no systemic issues of sample swapping or mislabeling. To ensure that only high-confidence samples are utilized in downstream analyses, we performed further quality control to filter participants as follows. Three samples displayed unusually high read-count variability at loci with heterozygous genotype calls as well as unusually low counts of high-quality variants, indicating that the sequenced sample may have been derived from a mixture of DNA of two individuals. Additionally, visualization of participant read coverage in sex chromosomes revealed four participants with potential aneuploidies. These seven participants were excluded from the WGS dataset. Thus, a sex assignment of “female” was interpreted as “XX” and “male” as “XY”. To mitigate the confounding effects of unknown familial relationships within the cohort, relatedness was assessed using KING 2.2.5, and an additional 120 participants were excluded to ensure the dataset contained no second-degree or closer relatives. This analysis also revealed three pairs of duplicate biological samples for which true participant identities were unclear. The three participants initially retained during relatedness filtering were thus excluded to arrive at the final count of 4,607 participants with WGS data after quality control and filtering. After haplotype phasing using SHAPEIT 4.2.1 13 , RFMix version 2 14 was used to estimate local ancestry information from WGS data for 29,062,806 harmonized variants in the 4,607 PEGS participants remaining after quality control. Variants genotyped in both the reference panel and PEGS were considered. The reference population for haplotype estimation and local ancestry inference was 2,504 unrelated individuals from the 1000 Genomes Project, Phase 3, with sequencing to a targeted depth of 30×  15 . Based on 1000 Genomes, each genomic position was classified by five continental super-populations: AFR (African), AMR (Admixed American), EAS (East Asian), EUR (European), and SAS (South Asian) 15 . Each genomic position had two assignments, one for each strand. Genome-wide ancestral similarity proportions were calculated for each participant based on the 29,062,806 assignments. Polygenic scores (PGS) for PEGS participants with WGS data were computed using data and metadata from the Polygenic Score Catalog 16 , 17 . To maximize the variants available for PGS calculation and ensure homogeneity across participants, we performed genome-wide imputation via the TOPMed imputation server 18 (TIS) using Minimac4 and the TOPMed r2 reference panel. After quality control, 4,600 participants with 2,553,563 variants had eligible data for imputation after removing indels, variants with a call rate < 90%, and variants excluded based on the McCarthy Group tools using the TOPMed Freeze 5 reference panel and pruning to LD r 2   0.3 were retained for analysis. PGS and their metadata were downloaded from the Polygenic Score Catalog. Data were harmonized using measures of variant and allele harmonization, including assignment of rsID (where possible) from the PGS Catalog, to allow the creation of files for builds 37 and 38 (via liftover). The PGS Catalog model submission requires the following variant-related fields: chromosome, position, effect allele, and weight, and the non-effect (or “other”) allele is “strongly recommended”. Absence of the “other” allele field can lead to ambiguity with respect to strand and/or multiallelic variants and thus the PGS Catalog data harmonization process attempts to augment this information when missing. To avoid the issue of inconsistency in encoding the number of risk variants on the X chromosome for males vs. females, we restricted to autosomes for PGS calculations. PGS were computed for PEGS participants from each PGS model using the PLINK 2.0 –score function 19 . Prior to computing scores, we removed palindromic (i.e., A/T or C/G) SNPs and harmonized the PEGS genetic data to match the PGS model effect allele. For each PGS model, we retained the total number of variants in the model, the number excluded, and the number used to create the PGS. Of the 3,490 scores in the PGS catalog (as of May 2023), we excluded 460 scores because less than 70% of the genetic variants were available in the imputed genotype data for PEGS participants. Thus, we were able to compute 3,030 PGS for 4,600 participants. A detailed description of PGS computation is provided in Schaid et al . 20 . Genome-wide methylation profiling was collected using the Infinium MethylationEPIC v1.0 BeadChip Kit for 4,724 PEGS participants with available WGS data. The EPIC chip targets 866,297 CpG sites in the most biologically significant regions of the human methylome. After quality control for samples and CpG sites, DNA methylation profile data were available for 826,286 CpG sites for 4,260 participants. Genomic DNA was extracted from aliquots of whole blood using an automated system (Autopure LS, Gentra Systems) in the NIEHS Molecular Genetics Core Facility or using DNAQuik at BioServe Biotechnologies Ltd. (Beltsville, MD). One microgram of DNA from each participant was bisulfite-converted in 96-well plates using the EZ DNA Methylation Kit (Zymo Research, Orange County, CA), and methylation analysis was conducted at the NCI Center for Cancer Research genomics core center. Samples were tested for completion of bisulfite conversion, and converted DNA was analyzed on Illumina Human MethylationEPIC arrays following the manufacturer’s protocol. The arrays were analyzed with high-throughput robotics to minimize batch effects. R software package ENmix was employed to perform data quality control and preprocessing, including ENmix background correction, RELIC dye bias correction, separate quantile normalization on methylated and unmethylated intensity values for type I and type II probes, and regression on correlated probes (RCP) for probe type bias adjustment. A total of 480 samples were excluded because of a low-quality CpG value > 5% or bisulfite intensity value 5% low-quality data were excluded. Low-quality data were defined as detection P value > 10 −6 or number of beads < 3. In 2020, geocoding was performed by assigning geographic coordinates to participant addresses, enabling proximity and grid-based analysis using distance to contaminant sources and area-level air pollutant concentrations as surrogates for exposure. For each participant, mapping was completed for five participant-provided addresses: at initial enrollment, at completion of the Health and Exposure Survey, at completion of the External Exposome Survey, the longest-lived childhood address, and the longest-lived adulthood address from the External Exposome Survey. The proximity of participants’ geocoded addresses to a variety of multi-pollutant point sources was calculated using publicly available data from federal and state regulatory agencies and then used to estimate exposure indicators. Figure  7 is an overview of the point sources used to estimate exposure indicators based on the geocoded addresses. Geocoding was performed for five participant-provided addresses: (1) address at initial enrollment; (2) address at completion of the Health and Exposure Survey; (3) address at completion of the External Exposome Survey; (4) the longest-lived childhood address; and (5) the longest-lived adulthood address. CDC/ATSDR Social Vulnerability Index (SVI) ( https://www.atsdr.cdc.gov/placeandhealth/svi/index.html ) and Environmental Justice Index (EJI) ( https://www.atsdr.cdc.gov/placeandhealth/eji/index.html ) values, which summarize multiple social determinants of health (SDOH) and measures of the impacts of environmental injustice on health, were linked at the census-tract level to each geocoded address using the SVI/EJI release year corresponding to the closest available time point. For historical addresses, contemporary SVI/EJI values were used due to limited historical index availability, which may introduce temporal misclassification. Participants’ ambient environmental exposures were estimated by linking publicly available environmental quality data to their geocoded residential addresses. Geocoding was performed using a three-stage process involving 1) U.S. Census Bureau Geocoding Services, 2) the Texas A&M Geocoding Service, and 3) manual review using Quantum GIS referencing public U.S. Census Bureau TIGER/Line street and Landsat orthoimage data. After cleaning and standardizing addresses (e.g., resolving spelling issues and parsing address components), the U.S. Census Bureau Geocoding service performed a first geocoding pass. Failed geocodes were then submitted to the Texas A&M Geocoding Service, which uses a sophisticated fuzzy-matching algorithm to match addresses typed with errors. The remaining unmatched addresses were geocoded using the manual review process, and a random sample of both hand-geocoded and algorithmically geocoded records was re-coded by a second human geocoder for quality assurance. Table  5 summarizes the geospatial data available for the PEGS cohort, and Table  6 outlines the available geocoding and GIS data linkages. Table 6 Summary of PEGS geospatial data. Source Description Examples Geocodes (GIS) Geocoded data from multiple participant-provided addresses at time of initial enrollment, completion of the Health and Exposure Survey, completion of the External Exposome Survey, the longest-lived childhood address, and the longest-lived adulthood address from the External Exposome Survey Geographic coordinates (latitude and longitude) from multiple participant-provided addresses Hazards Exposure estimates computed from Department of Transportation (DOT) data Information from train tracks, rail depots, and roadways such as total major roadway length and distance to the nearest rail depot Hazards Exposure estimates computed from Federal Aviation Administration (FAA) data Information from aircraft departure and arrival sites (e.g., distance to the nearest airport) Hazards Exposure estimates computed from Federal Communications Commission (FCC) data Information from cellular network towers (e.g., nearest cell tower) Hazards Exposure estimates computed from North Carolina Department of Environmental Quality (NCDEQ) Distance to multi-pollutant point sources such as swine caged feeding operations (CAFOs), hazardous waste sites, hazardous spill sites, EPA superfund sites, and wastewater treatment plant release sites Hazards Exposure estimates computed from Nuclear Regulatory Commission (NRC) data Distance to a nuclear power station Hazards Exposure estimates computed from Atmospheric Composition and Analysis Group (ACAG) data Particulate matter concentrations such as PM2.5 total, PM2.5 sulfate, PM2.5 black carbon, and other Hazards Exposure estimates computed from Center for Air, Climate, and Energy Solutions (CACES) data Concentrations for multiple pollutants such as carbon monoxide, nitrogen dioxide, and ozone concentration Hazards Exposure estimates computed from Toxics Release Inventory (TRI) data Emissions for chemicals of interest such as benzene, ethylbenzene, xylene, and toluene MERRA-2 data (Earthdata) Geospatial data linkages from the Modern Era Retrospective Analysis for Research and Applications (MERRA-2) project to assimilate a range of satellite-based environmental observations into a consistent estimate of climate and environmental metrics Particulate, gas, meteorological, and health-relevant exposure indicators such as dust sedimentation, organic carbon emission bin, SO 2 biomass burning emissions, and sea-level pressure Social Vulnerability Index (SVI) Geospatial data linkages for CDC/ATSDR SVI designed to consistently quantify multiple social determinants of health across the United States over time Consists of summaries of social determinants of health at the census-tract level including an overall index, four component indexes (socioeconomic status, household characteristics, racial and ethnic minority status, and housing type/transportation), and source variables used to compute each index component (e.g., poverty, education, overcrowding, access to a vehicle) Environmental Justice Index (EJI) Geospatial data linkages for CDC/ATSDR EJI containing summaries and ranks of the cumulative impacts of environmental injustice on health at the census-tract level Ranks for each census tract on 36 environmental, social, and health factors grouped into 10 domains and three overarching modules – environmental burden, social vulnerability, and health vulnerability Overview of the geocoding and GIS data linkages available for the PEGS cohort. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Summary of PEGS geospatial data. Overview of the geocoding and GIS data linkages available for the PEGS cohort. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. GIS data linkages were performed using these geocodes and R version 4.0 and Geometry Engine Open Source (GEOS) bindings provided by the sp and sf R packages. For proximity measures (e.g., distance to the nearest hazard), we calculated great-circle distances in meters using the WGS84 spheroid. For point-in-polygon linkages (e.g., linking geocodes to measures computed for census tracts), participants were assigned the value of the containing polygons (e.g., census tract or grid cell) using GEOS spatial indexing in the WGS84 coordinate system. Temporal alignments were chosen based on the availability and richness of source data to support expected study designs. For example, land-use models with many available time points (e.g., Center for Air, Climate, and Energy Solutions (CACES) data) were simplified by averaging the containing grid cells over a long period. For area-based measures (e.g., the SVI) for which few discrete time referents are available, all available time points were linked to each geocoded address. De-identification procedures followed the NIST SP 800-122 Guide to Protecting the Confidentiality of Personally Identifiable Information (PII). All direct identifiers were removed prior to data sharing. Dates were generalized to year-only format, and residential address data were converted to geocoded coordinates for internal linkage and then removed from shared datasets. Geographic information provided to approved users is limited to derived exposure metrics and census-tract–level variables. Shared datasets were reviewed to ensure compliance with HIPAA Safe Harbor and Expert Determination standards to minimize re-identification risk.

A growing number of projects have used PEGS data. Data have been used to build predictive models for type 2 diabetes 21 , discover gene-environment interactions with immune-mediated disorders 22 , examine exposome-wide associations for cardiovascular diseases 23 , conduct association mapping of the HLA region in asthma 24 , and examine air pollution mixtures and autoimmune skin disease risk 25 . Additionally, several studies have called back participants to collect data to evaluate specific hypotheses. From its inception, a goal of PEGS has been to assist scientists and other investigators by identifying cases and controls for call-back and additional interviews or in-person studies. This has enabled studies investigating the effects of aging and ultraviolet light on somatic mutations that facilitate genome instability and carcinogenesis 26 . Further, these data have been combined with cell-based and mouse studies to uncover the role of a key innate immune receptor in asthma 27 .

Genetic and environmental factors interact in complex ways to influence individual-level health. Due to this intricate interplay, collecting and analyzing data on genetics and the environment are crucial for a comprehensive understanding of disease etiology and prevention tactics. Genetic data enables the identification of variations in an individual’s DNA that can affect disease susceptibility, providing insights into inherent risk factors, enabling personalized risk assessments, and informing early interventions. Combined with genetic data, environmental data is used to identify how genetic variation interacts with exposures, engendering a deeper understanding of how risk factors vary based on individual exposures. Globally, an estimated 25% to 90% of overall disease etiology can be attributed to environmental factors 1 – 5 . Moreover, many diseases result from interactions between an individual’s genetic predisposition and their particular environmental exposures such as smoking or air pollution, highlighting the importance of considering both genetic and environmental factors. Integrating genetic and environmental data enables the development of personalized treatment plans and allows medications to be tailored to individual genetic profiles, leading to safer, more effective treatments with reduced adverse effects. Additionally, by understanding individual genetic and environmental risk factors, healthcare providers can offer personalized lifestyle change recommendations and suggest interventions aimed to prevent or mitigate disease risk. While there is growing appreciation of the importance of including both genetic and environmental data in studies, the dearth of relevant datasets limits research opportunities. While whole-genome sequencing (WGS) data are available, collecting analogous exposome data is challenging. While exposome data are essential for elucidating the etiology of common, complex diseases, they are difficult to measure, collect, and compare. In this study, we present the Personalized Environment and Genes Study (PEGS), a unique data resource comprising genetic and extensive environmental exposure data available to the research community.

PEGS collects survey-based exposomic, genomic, and geographic information system (GIS) data from a racially and ethnically diverse North Carolina-based cohort of nearly 20,000 individuals. The PEGS website summarizes the study and describes the available data ( https://www.niehs.nih.gov/research/atniehs/labs/crb/studies/pegs/index.cfm ). While most studies focus on a single disease or environmental exposure, PEGS collects data on multiple diseases and environmental exposures, including diet and lifestyle factors, in concert with genetic data. The goal of PEGS is to integrate these large-scale, multi-dimensional data to enable researchers to dissect the etiology of disease and identify the collective effects of environment, diet, lifestyle, and genetic factors on human health. Originally established in 2002 for ongoing research at the National Institute of Environmental Health Sciences (NIEHS), the initial cross-sectional cohort of the Environmental Polymorphisms Registry (EPR) recruited participants by convenience sampling at community events. The EPR was renamed the Personalized Environment and Genes Study in 2022. Figure  1 is a timeline of the evolution of PEGS. PEGS represents a long-term investment by NIEHS in collecting and analyzing exposome data. The cohort has been expanded to function as an extensive repository of data on medication, lifestyle, and environmental exposures, diseases, and genetics, as shown in Fig.  2 . Participants complete three surveys, the Health and Exposure Survey and the Internal and External Exposome Surveys, that request information on exposures at home and work and lifestyle factors such as sleep and diet. PEGS participants provide broad consent to share their data with researchers and data repositories for all relevant research. The National Institutes of Health Institutional Review Board (NIH IRB #: 04E0053) has approved the study protocol, informed consent forms, recruitment materials, all participant materials, and human data sharing, including genomic data sharing and publications. PEGS participants also consent to provide biological samples taken from multiple sites, including blood, skin, cheek, and nasal cells, exhaled breath, hair, nail clippings, saliva, sperm, sputum, stool, baby teeth, urine, and household dust. There is an option for participant call-back to collect additional tissue samples for specific studies with participant consent, including those collected during medical treatment or saved after biopsy or surgery. Fig. 1 Timeline of the progression and expanded scope of the Personalized Environment and Genes Study (PEGS). Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Fig. 2 Overview of available PEGS data. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Timeline of the progression and expanded scope of the Personalized Environment and Genes Study (PEGS). Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Overview of available PEGS data. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. The 19,445 PEGS participants are demographically representative of the population of North Carolina. Table  1 provides demographic details, and Fig.  3 is a visual overview. Roughly two-thirds of participants are female (62.3%; 12,120), and one-third are male (37.6%; 7,298). Approximately two-thirds of participants self-identify as White (63%; 12,279), slightly over one-quarter self-identify as Black (27.6%; 5,361), and 4.5% self-identify as another race (868). Approximately 5% of participants self-identify as Hispanic (5.1%; 979). The cohort includes participants of varying education level, socioeconomic status, and age (range: 18.4–98.3 years), with a mean age of 50.2 years (at completion of the Health and Exposure Survey). Due to the diversity of the cohort, researchers can investigate disease risk in multiple populations and uncover health disparities across groups resulting from disproportionate environmental exposures. As a result, the findings from research using PEGS data are broadly applicable. Table 1 PEGS participant demographics. Sex (N, %) Female 12120 (62.3) Male 7298 (37.5) Race (N, %) White 12279 (63.1) Black or African American 5361 (27.6) Other 868 (4.5) Ethnicity (N, %) Non-Hispanic/Non-Latino 17750 (91.3) Hispanic/Latino 979 (5.0) Education (N, %) Grade 12 or less 1563 (16.5) College, technical, or vocational 2794 (29.6) Bachelor’s degree 2552 (27.0) Graduate or professional degree 2469 (26.1) Income (N, %) Less than $20,000 1294 (13.7) $20,000 to $49,999 2815 (29.8) $50,000 to $79,999 2251 (23.8) $80,000 or more 2758 (29.2) Age (mean, SD) 50.2 (16.0) Demographics of participants in the PEGS cohort. Age is at completion of the Health and Exposure Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Fig. 3 Demographics for PEGS participants, showing age at completion of the Health and Exposure Survey and self-reported sex, race, education level, and income level. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. PEGS participant demographics. Demographics of participants in the PEGS cohort. Age is at completion of the Health and Exposure Survey. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. Demographics for PEGS participants, showing age at completion of the Health and Exposure Survey and self-reported sex, race, education level, and income level. Data reported from PEGS Data Freeze 3.1 created on 6/27/2023. PEGS participants are recruited in outpatient clinics and hospitals and at over 200 unique sites in North Carolina that include businesses, career fairs, universities, community events such as charity walks, health fairs and festivals, senior centers, sporting events, social clubs, religious institutions, and military bases. Enrollment for the PEGS cohort is ongoing at the NIEHS Clinical Research Unit ( https://www.niehs.nih.gov/research/atniehs/labs/crb/durham/index.cfm ) in North Carolina, the PEGS enrollment website ( https://joinastudy.niehs.nih.gov/studies/pegs ), and Clinicaltrials.gov (ClinicalTrials.gov ID: NCT00341237 ). Contact is maintained with over 6,000 PEGS participants. An important aspect of the cohort is the availability to recall participants to support additional data collection for follow-up studies to enable validation efforts, multi-omics data collection, and add-on studies. Through a collaboration with the National Human Genome Institute (NHGRI) Reverse Phenotyping Core, which has collected de-identified exome data from multiple investigators in the NIH intramural research program, researchers can contact participants for studies to determine the influence of genotypes of interest on individual phenotypes. To protect the privacy, safety, and confidentiality of PEGS participants, PEGS data have not been deposited in a publicly available repository. The data are securely stored in a centralized, shared repository at NIEHS. To ensure the data are safely and securely available for research collaborations, de-identified versions of PEGS Data Freezes without personally identifiable information (PII) have been created for use when sharing PII or protected health information (PHI) is inappropriate and/or not required. PII or PHI was removed from all PEGS data components following the guidelines specified in the “NIST SP 800-122 Guide to protecting the confidentiality of personally identifiable information (PII)” by McCallister et al . 6 . Briefly, dates more specific than a year, participant address information more specific than state or territory, and participant identifiers that could be linked to health records or patient information were removed.