How to evlauate the result: GenotypeConcordance or Concordance

Answered

Follow

lizhichao

• February 17, 2022 09:09

REQUIRED for all errors and issues:
a) GATK version used:gatk-4.1.7 
b) Exact command used:/zfssz2/ST_MCHRI/BIGDATA/USER/liupanhong/Src/gatk-4.1.7.0/gatk --java-options -Xmx3G GenotypeConcordance \
--CALL_VCF $testvcf \
--OUTPUT ${sample}_zbolt_microarray \
--IGNORE_FILTER_STATUS \
--TRUTH_VCF $baseline  \
--TRUTH_SAMPLE $sample && echo "${sample} done" 

c) Entire program log:

when I analysis some samples with different analysis pipeline,  I want to know which one performs more precise, so I will run rtg eval or hap.py .

Besides, when I gonna to get the consistency,which one is ok between GenotypeConcordance and  Concordance?
See forum topic details at forum guidelines page: https://gatk.broadinstitute.org/hc/en-us/articles/360053845952-Forum-Guidelines

0

• 

  Genevieve Brandt (she/her)

  • February 22, 2022 21:10

  Hi lizhichao,

  The tools are very similar but GenotypeConcordance calculates the concordance at the sample genotype level while Concordance calculates the concordance at the site level.

  The one you choose depends on the information you want for further analysis.

  Best,

  Genevieve

  1

  Comment actions Permalink
• 

  lizhichao

  • March 02, 2022 11:12

  Sorry ，I cant get the difference between the genotype level and site level. By the way ,I did noticed that some paper calculate the precision at variant calls and  genotype level .But I think the genotype is based on every site/variant, so I cant figure that .Can you explain it to me? thanks!

  0

  Comment actions Permalink
• 

  Genevieve Brandt (she/her)

  • March 02, 2022 19:45

  Yes! I can explain this. 

  In a multi-sample VCF, there is some information in the VCF that applies to all samples and some information that applies to each specific sample. The site level calculation refers to a calculation from values that apply to all samples at that site, so data in the INFO field or REF/ALT fields. 

  A genotype level calculation is looking at the FORMAT field for each sample individual and calculating the concordance with those GT, AD, AF, etc, values.

  There is an in depth explanation in our VCF document, which you can see here: https://gatk.broadinstitute.org/hc/en-us/articles/360035531692-VCF-Variant-Call-Format

  Hope this helps!

  1

  Comment actions Permalink
• 

  lizhichao

  • March 03, 2022 01:22

  if I compare two sample from two individual vcfs, then the above explanation is suited for that? Does the Site level calculate the REF/ATL  difference of overlap sites of the two vcfs? the Genotype level calculates the GT difference of overlap sites?

  0

  Comment actions Permalink
• 

  Genevieve Brandt (she/her)

  • March 04, 2022 02:33

  Yes, that's correct!

  1

  Comment actions Permalink
• 

  lizhichao

  • March 09, 2022 04:26

  Thanks for your answer!

  0

  Comment actions Permalink

Please sign in to leave a comment.