This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Converting string to numerical in bcftools

Hi everyone,

I am using bcftools to filter variants from a VCF file. The variants from this VCF file have been annotated using ANNOVAR.

I would like to filter variants having a CADD score > 20 in a field named "CADD_phred" which has been created and annotated by ANNOVAR.

However, when the following command does not work:

bcftools view -i'CADD_phred>20' -Oz -o out.vcf.gz in.vcf.gz

I believe it is because the CADD_phred information has been stored as a string and does not contain numerical value. For example, I was able to remove the variants without CADD scores (stored as ".") using the bcftools view -e'CADD_phred="."' command.

Is there any solution to convert string values to numerical values using bcftools?

Thanks!

bcftools
• 3,253 views • link
updated by AMARU • AMARU • written by pjuge • pjuge

Hi,

I was wondering if you were able to fix that, I have the exact same issue. This is my case, it is a string, I changed to float but wasnt able to filter by CADD>20.

##INFO=<ID=CADD_phred,Number=.,Type=String,Description="CADD_phred annotation provided by ANNOVAR">

Any thoughts?

Thanks

• 0 views • link
updated by GenoMax GenoMax • written by AMARU • AMARU

1 answer

I'm not sure of a simple single command to convert string to integer. However, you could do it through a few steps.

  1. Print the full header from the VCF to a new file using bcftools view -h > header.txt
  2. Modify the header either in a text editor or using something like sed to change the info for CADD_phred to 'Type=Integer'
  3. Reheader the VCF file using bcftools reheader -h header.txt -o new_header.vcf

I assume that will work, but hard to say without seeing your actual VCF file.

• 0 views • link
written by dr_cfoster dr_cfoster

Thank you for your reply! I was able to change the header line

##INFO=<ID=CADD_phred,Number=.,Type=String,Description="CADD_phred annotation provided by ANNOVAR">

to

##INFO=<ID=CADD_phred,Number=.,Type=Float,Description="CADD_phred annotation provided by ANNOVAR">

without any effect on the filtering by CADD_phred>20

As I noticed that other information containing numbers were indicated as "Number=1", I also tried to change it to

##INFO=<ID=CADD_phred,Number=1,Type=Float,Description="CADD_phred annotation provided by ANNOVAR">

without effect...

• 0 views • link
updated by GenoMax GenoMax • written by pjuge • pjuge

Make sure that "CADD_phred" is actually a float and not an integer. Also, when filtering with bcftools view, try explicitly indicating that "CADD_phred" is defined as an "INFO" tag: bcftools view -i "INFO/CADD_phred>20" -Oz -o out.vcf.gz in.vcf.gz. If that doesn't work, provide an example truncated VCF file showing the header and at least one variant with a CADD_phred score.

• 0 views • link
written by dr_cfoster dr_cfoster

Following the above example, How can I make sure "CADD_phred" is float and not integer?

Thanks

• 0 views • link
written by AMARU • AMARU

As I said in the previous reply, I/we probably can't help without an example truncated VCF file showing the header and at least one variant with a CADD_phred score. Without this I/we are simply guessing at solutions.

Is the number associated with CADD_phred in the form of 123.45? If so, it's a float, and the header might need Type=Float. Is it in the form of 123? If so, it's an integer, and the header might need Type=Integer. These suggestions might even be completely down the wrong track, but, again, no way to know without basic example data.

• 0 views • link
written by dr_cfoster dr_cfoster

Hey. I just edited the CADD_score of the vcf file in a text editor. It worked. Thanks for commenting.

• 0 views • link
written by AMARU • AMARU

Log in to answer this question.

More posts like this