5 Annotation and Search for Causal Variants

5 Annotation and Search for Causal Variants

contact: ludwig@univ-brest.fr

Annotation is adding information from external sources:

  • Genes/Transcripts Impacted
  • Consequences
  • Phenotypical Information
  • Validation
  • Frequencies from different populations

There are Several Annotation Softwares:

5.1. Online Annotation and Filtering

Fill in the following fields:

Key Value
Species Homo_sapiens
Name for this job TP3.1_votre_nom
Input data, provide file URL https://lysine.univ-brest.fr/~tludwig/master2/onlyrare.grch38.variants.vcf
Filtering options Show one selected consequence per variant

Based on the table from https://m.ensembl.org/info/genome/variation/prediction/predicted_data.html, answer the following questions

consequences.png

You can answer the first 2 questions while your job is still Queued or Running

5.1.1. What are the different IMPACT classes? What do they mean?

5.1.2. What is the IMPACT class of missense_variant? Why not another one?

5.1.3. Use filters to answer

How many variants are there of the following type:

  1. synonymous_variant
  2. 3_prime_UTR_variant (use the edit button)
  3. frameshift_variant

Use Filters, Consequence, is, XXXX

5.1.4. Look at the consequence of missense_variant on the protein

  • Use Filters, Consequence, is, missense_variant
  • Click Show/hide columns and select SIFT,PolyPhen and AF
  • Sort by AF

5.2. Online Multiple Annotation and Filtering

Here, variants that are frequent in the general population have been filtered out, and only de novo variants have been kept. All non-exonic variants and variants that have no impact on the protein have also been filtered out. The studied pathology as severe intellectual deficiency.

Key Value
Species Homo_sapiens
Name for this job TP3.2_votre_nom
Input data, provide file URL http://lysine.univ-brest.fr/~tludwig/master2/2.denovo.grch38.vcf
Variants and frequency data deselect 1000 Genomes global minor allele frequency and select gnomAD (genomes) allele frequencies
Filtering options Show all results

  • Display only the following columns

    • Uploaded Variant
    • Location
    • Allele
    • Consequence
    • Symbol
    • Exon
    • Protein position
    • Amino Acids
    • Codons
    • SIFT
    • PolyPhen
    • gnomAD_AF

  • Sort by Location

5.2.1. How many variants are there?

Don't answer to fast 😉

5.2.2. How many are new?

That is unknown in dbSNP

5.2.3. How many missense_variants ?

Use filters

5.2.4. Which variant can be immediately eliminated as a candidate?

This pathology being extremely rare. Which variant can be immediately eliminated as a candidate?

5.2.5. What are the 2 variants with the most severe effect?

Based on SIFT/PolyPhen annotation. What variants have the most severe effect on the protein?

5.2.6. In what genes are these variants?

5.2.7. Which variant seems to be causing the disease? Why?

Read about the target genes found in the previous question.

5.2.8. In which exon of the gene is the variant located?

5.2.9. What is the variant's impact of the protein?

5.2.10. What is its effect on the protein?

According to SIFT and PolyPhen

5.2.11. What is the position of the variant?

5.2.12. How well is this amino acid conserved ?

Look for this position in UCSC browser.

5.3. Local Annotation and Filtering

Medical Information
  • Exome Sequencing
  • Severe Intellectual Deficiency

5.3.1. Annotation with variant effect predictor

Annotate the variants with vep

cd

### IF TOO SLOW
/DATA/bioinfo/annotate.sh

### REAL COMMAND
vep variants.vcf.gz annotated.variant.vcf

bash

5.3.2. Use VCFProcessor

Documentation : http://lysine.univ-brest.fr/vcfprocessor/

5.3.3. Keep only rare variants

Keeping only variants with an allele frequency < 1% in1kG, 1kG_Europe, gnomAD_exome and gnomAD_exome_europe

VCFProcessor FilterFrequencies \
	--vcf annotated.variant.vcf \
	--threshold 0.01 \
	--pop AF,EUR_AF,gnomADe_AF,gnomADe_NFE_AF\
	--out rare.variant.vcf

bash

How many variants (sites) are left ?

5.3.4. Keep only variants that strongly affect the protein

Of those variants, keeping only variants with an effect at least as important as splice_region_variant

VCFProcessor FilterConsequenceLevel \
    --vcf rare.variant.vcf \
    --csq splice_region_variant \
    --out effect.rare.variant.vcf

bash

How many variants (sites) are left ?

5.3.5. Keep only variants compatible with a recessive transmission pattern

VCFProcessor Recessive \
    --vcf effect.rare.variant.vcf \
    --ped /DATA/bioinfo/family.ped \
    --missing false \
    --nohomo true \
    --mode strict \
    --out recessive.effect.rare.variant.vcf

bash

How many variants (sites) are left ?

5.3.6. Keep only variants compatible with a de novo transmission pattern

VCFProcessor DeNovo \
    --vcf effect.rare.variant.vcf \
    --ped /DATA/bioinfo/family.ped \
    --missing false \
    --out denovo.effect.rare.variant.vcf

bash

How many variants (sites) are left ?

5.3.7. List of implied genes

Extracting the genes involved in the "recessive" and "de novo" variants

VCFProcessor GeneList \
    --vcf recessive.effect.rare.variant.vcf \
    --out recessive.genes.list

VCFProcessor GeneList \
    --vcf denovo.effect.rare.variant.vcf \
    --out denovo.genes.list

bash

5.3.8. Results

The last commands show the list of genes.

echo -n "Recessive "
cat recessive.genes.list
echo 
echo -n "de novo "
cat denovo.genes.list
echo 
echo "Union of target genes"
cat *.genes.list | sort -u | grep -v genes | grep -v "-" | grep -v "\."

bash

Which gene (and thus variant) seems to be the best candidate to explain the patient's phenotype ?