Test 2: A celebration of your new SNP calling expertise - mestato/EPP622_2026 GitHub Wiki

Note: Reads were simulated as diploid reads

I have taken the Saccharomyces cerevisiae genome and engineered in 5000 SNPs. Then I simulated 4 sets of Illumina reads from that genome. They vary in quality, level of adapter or poly-G contamination, and coverage. Your job is to

  • Use the pipeline from class labs to call SNPs from the original Saccharomyces cerevisiae genome. Your job is to find those 5000 snps.
  • Along the way you will find the quality issues in each sample and compare how those quality issues do or do not impact downstream snp calling.
  • Document your work on a github wiki, send in multiqc/stats reports from each step.
  • Interpret your results - how did quality impact final outcomes? How did filtering? I am providing a script that calculates recall, precision and F1 of your data aginst the truth set to help with this.
  • Finally, compare the class pipeline to the pipeline from your AI SNP calling exercise that you turned in as homework. I will be providing some specific questions to think about and answer.

You will need to set up a github wiki template. To get set up:

  • Register a github account if you don't already have one
  • Create a new repository called "EPP622-2026-SNPCalling"
  • Go to the wiki tab at the top and start your documentation there.
  • Github wikis use the markdown language for formatting, learn more here

Individual Work Expected: You may discuss concepts with classmates; the commands, wiki, and writing must be your own. Do not share code.

Using AI for this test: You may use AI assistants. The last part, the analysis of the AI pipeline, requires it. But the wiki must document work you performed. You may be asked, in person, to walk through the steps of your pipeline and explain your choices and interpret the results.

October 13th - Optional deadline to turn in work for Steps 1-6, from fastqc to evaluating your SNP sets against the ground truth set. Zane and I will help you figure out if you have any issues to fix prior to answer the final two sections, i.e. the questions asking you to interpret your results and to think about AI usage.

October 20th - Final due date

Files

  • Reference genome (sacCer3.fa) plus many of the indices you will need /lustre/isaac24/proj/UTK0505/test2/ref
  • Read 1 and read 2 for samples A, B, C and D /lustre/isaac24/proj/UTK0505/test2/reads
  • Python script (evaluate_snps_v4.py) and truth set (truth_snps.vcf) to check your SNP calls against the ground truth (more description on how to use these is available in the wiki template.) /lustre/isaac24/proj/UTK0505/test2/check
  • Test 2 Wiki template rendered
  • Test 2 Wiki template raw

Rubric

1. Wiki documentation (30%) Goals: assessing command line skill development; reproducibility, documentation and organization skills. I expect sensible formatting of wiki (use headers, formatted code blocks) and sensible formatting of work on the server filesystem (subfolders for each analysis, good file naming).

  • Step 1 - Fastqc comments, commands, and results following template
  • Step 2 - Fastp comments, commands, and results following template
  • Step 3 - bwa-mem2 comments, commands, and results following template
  • Step 4 - bcftools snp calling and genotyping comments, commands, and results following template
  • Step 5 - bcftools filtering comments, commands, and results following template
  • Step 6 - evaluation of your calls against the ground truth - comments, commands, and results following template
  • directory structure - see template

2. MultiQC and stats reports (30%) Goals: assessing command line development, understanding of pipeline mechanics, ability to identify incorrect or incomplete results

Provide paths to these files to instructor and TA:

  • Fastqc - pretrimming multiqc
  • Fastqc - posttrimming multiqc
  • bwa-mem2 multiqc (from samtools stats, samtools flagstats, and samtools coverage)
  • output of bcftools stats for raw, gentle and stringent snp sets

3. Questions: Assignment Analysis (20%)

Goals: assessing understanding of SNP calling concepts and ability to evaluate the meaning of results of the pipeline.

See specifics and embed your writing in the wiki markdown.

4. Questions: AI Analysis (20%)

Goals: assessing understanding of SNP calling concepts and ability to evaluate the use the AI for executing bioinformatic tasks. Taking a pro-AI, an anti-AI or an ambivalent/nuaced ("bit of both") stance is fine. I'm looking for thoughtful engagement with how you could or would use AI and ensure your results are correct.

See specifics and embed your writing in the wiki markdown.