45 Matching Annotations
  1. Last 7 days
  2. localhost:8000 localhost:8000
    Anthropic团队如何使用Claude Code - 详细案例研究
    45
    15. chrizelle_kok 15 Oct 2026
      2 Variant Filtering

      Transition to Transversion Ratio

      The transition to transversion ratio (Ti/Tv) compares the number of transition mutations to the number of transversion mutations observed within a sequence. Due to there being twice as many possible transversions to transitions, we would expect a Ti/Tv of 0.5. However, this changes when we take into consideration biological context. In human genomes, this expected Ti/Tv can range anywhere from ~2.0 to ~3.3. The ratio is also highly depenedant on genomic region, ancestry, analysis pipelines, filtering etc. As a QC, it can be used to evaluate the quality of a sequence of SNPs for technical errors. A Ti/Tv closer to the theroretical 0.5 value can indicate sequencing errors (as these errors mimic random mutation rates) and extremely high Ti/Tv values (>4) can indicate high bias in analysis. [@gatk2024qc; @cdgenomics2026; @wang2015; @carson2014]

      Mutational Spectrum Plot

      Here, we break down all mutations in a sample according to each type of mutation. There are six possible mutations. [@griffithlab2017]

      References:

      Broad Institute GATK Documentation: https://gatk.broadinstitute.org/hc/en-us/articles/360035531572-Evaluating-the-quality-of-a-germline-short-variant-callset

      CD Genomics Technical Resource: https://www.cd-genomics.com/resource-quality-control-whole-exome-sequencing.html

      PMID: 25297068, 24884706

      Principal Component Analysis (PCA)

      PCA is a statistical method used to simplify high-dimensional genetic data (such as millions of SNPs) into a small number of continuous axes called principal components (PCs). This is useful to infer ancestry and investigate population structure. It can also be used to check for batch effects in your dataset. [@wickland2021]

      References:

      PMID: 33861770

      Addition References
    31. chrizelle_kok 12 Oct 2026
      Figure 15: Homozygous Reference

      I think we need to add a "Notes" section detailing the thresholds like in section 1. Something like:

      "We set the cutoff to be ##, i.e. values above and/or below this cutoff is likely to increase your false positive rate"

    32. chrizelle_kok 12 Oct 2026
      Lorem

      The Allele Balance is the ratio of sequence reads supporting a specific allele compared to the total number of reads covering that genomic position, i.e. it measures the proportion of sequencing reads supporting a specific allele relative to the total reads at a given locus for a single sample. Ideally, you want the AB ratio of your heterozygous loci (0/1) to be ~0.5 (ideal: 50% ref, 50% alt reads) and the homozygous loci (reference 0/0 and alternate 1/1) ~0.0 (Ideal: 0% alt reads) and ~1.0 (Ideal: 100% alt reads) respectively. However, what we see in practice is that true variants cluster around these discrete theoretical ratios. In filtering pipelines, calls deviating significantly from the theoretical thresholds are typically filtered out as false positives caused by strand bias, PCR duplication, or misaligned reads.

      The second quality evaluated in this filter is GQ and AD. The specifics of these parameters are detailed in section 1.1. You will se that we plot the AB vs GQ and AD.

      Variants that fail to pass all three metrics will be removed.

      Addition
    35. chrizelle_kok 12 Oct 2026
      Lorem

      QC has now been done. In this section, we will now filter the variants based on functional and quality information. These filters are designed to enrich for rare variants with a predicted functional impact, i.e. a variant that has some evidence for pathogenicity and biological relevance.

      Addition
    38. chrizelle_kok 12 Oct 2026
      we have no confidence in any genotype with an AF

      I think more appropriate wording would be: "we have low confidence that an allele is disease-causing if it is present in more then 5% of the population (AF above 0.05)"

      Thoughts
    42. chrizelle_kok 12 Oct 2026
      filtering.

      I think we should consider adding an explanation on the difference between entries and rows here. So maybe add something like "In this workflow, we make use of a data structure called a hail MatrixTable. In the following filters, you will see that variants retained and removed are calculated and displayed by rows and entries. Entries represent single variants in single samples, i.e. a single variant. Rows represent a variant across all samples, i.e. the same variant in all samples."

      Thoughts
    45. chrizelle_kok 12 Oct 2026
      VARIANT QUALITY CHECKS AND FILTERING REPORT

      List of document changes that need to be made: * Review current text * Add missing text * Add references to text

      List of additional/changed visualisations (code) * Add PCA to (1) * Add Ti/Tv Ratio to (1) * Add sunburst charts to (2)

    Visit annotations in context

    Tags

    Annotators

    URL

    localhost:8000/main.html