AI4All_Genomics

AI4All@Princeton is a summer camp that aims to promote diversity in computer science by teaching AI to young students of diverse backgrounds (https://ai4all.princeton.edu). In this module, we will be investigating our genomic diversity by exploring natural genomic variation between world populations.

Relevant Links:

• IGSR home: Genohttp://www.internationalgenome.org/home
• Phase 3 1000 Genomes Results: https://www.nature.com/articles/nature15393#abstract
• Phase 3 1000 Genomes Data: https://ftp-trace.ncbi.nih.gov/1000genomes/ftp/release/20130502/
• Phase 3 1000 Genomes Processing Example: https://bitbucket.org/remills/1000gp_sv_phase3/src/master/
  • *** see gwas_sv_ld_filt_af.txt

Data Preprocessing

• Run GATK software to subset SNPs: https://software.broadinstitute.org/gatk/documentation/tooldocs/4.0.0.0/org_broadinstitute_hellbender_tools_walkers_variantutils_SelectVariants.php
  • run ./gatk/gatk SelectVariants -V input.vcf -O output.vcf --keep-ids gwas_sv_ld_RSIDs.list
    • example output.vcf: chrX_filtered.txt
• concatenate vcf files using vcf-tools
• https://vcftools.github.io/man_latest.html

Input Data:

• "chr01-22_filtered.vcf"
• to download: https://drive.google.com/drive/folders/1O7cRyGbEHrkjiCAkaKODN0V2HYTDXjss

To Do:

• Outline first two weeks

• Implement first two weeks notebooks (goals)

• start first 3 mini lectures

• plan (and test) which ML algorithms to introduce to students (clustering, standard prediction, etc)

• brainstorm intro slides to SNPs

• explore with python and benchmark runtimes

Files:

• 1000genomes_dataExploration.ipynb: preliminary exploration of 1000 Genomes data (PCA, SVM, data cleanup and filtering)