AD_STR

Repository for code used in Guo et al. analysis of short tandem repeats in Alzheimer's disease.

STR genotyping is performed using gangSTR v2.4.0 (available at https://github.com/gymreklab/GangSTR) and ExpansionHunter v5 (https://github.com/Illumina/ExpansionHunter). Both software programs were run on Amazon Web Services. The approximate run time using 4 Gb of RAM on AWS is 8 hours per sample for each software program.

Custom code for processing STR genotype data was performed using R v3.6.3. The software depends on the following publicly available packages in R. Approximate installation time is < 5 minutes on a MacBook Pro.

• regioneR v1.30.0
• clusterProfiler v4.2.2
• GenomicRanges v1.38.0
• dbscan v1.1-11
• RNOmni v1.0.1
• stringr v1.5.0
• reshape2 v1.4.4
• data.table v1.13.0
• RNOmni v1.0.1

Here is a summary of the code: A) STR genotyping

1. ExpansionHunter genotyping (run_EH.sh)
2. gangSTR genotying (run_gangSTR.sh)
3. ExpnasionHunter json file (eh.v5_w_gangstr.v13.polymorphic.json.gz)
4. gangSTR STR bed file (gangstr.v13.polymorphic_w_eh.v5_offtarget.exp10_p05.top5_no_segdup.fast_2s.bed.gz)
5. Code to extract coverage from ExpansionHunter vcf files (parse_eh_coverage.R)
6. Code to extract STR genotypes from ExpansionHunter vcf files (parse_eh_genotypes.R)
7. Code to extract STR genotypes from gangSTR vcf files (parse_gangstr_genotypes.R)
8. Code to extract coverage from gangSTR vcf files (parse_gangstr_coverage.R)

B) Single STR association analyses (single_str_association_analysis.R)

C) Fisher's exact test for single STR burden analysis (fisher_single_str_test.R)

D) STR expansion analysis

1. Identification of STR expansions using DBSCAN algorithm (run_dbscan.R)
2. Parse DBSCAN results (parse_dbscan.R)

E) Genomic annotations and enrichments

1. Perform genomic annotations
2. Run chromHMM enrichments (run_chromhmm_enrichment.R)
3. Run transposable element enrichments (run_te_enrichment.R)
4. Run Hippocampus histone ChIP enrichments from ENCODE (run_encode_histone_chip_enrichment.R)

We also provide a simulated dataset for trialing the DBSCAN algorithm. This is provided along with a README in the test_dbscan folder.