Somatic indel calling with Pindel

Question

[sichan]

Hello,

I noticed that Pindel was previously reported to not be able to call somatic indels without post-processing, e.g. see these posts:

• Pindel Somatic Calls?
• http://seqanswers.com/forums/showthread.php?t=42881

However, I noticed that the latest version of Pindel (0.2.5b1) has the --NormalSamples option. According to the usage prompt, this flag turns on germline variant filtering:

-N/--NormalSamples
Turn on germline filtering, less sensistive and you may miss somatic calls (default false)

I run Pindel as follows:

pindel -f hg19.fa -i bam_config_file.txt -o Test1 --NormalSamples

My bam configuration file looks like:

/path/to/tumor/bam insert_size Tumor
/path/to/normal/bam insert_size Normal

However, I noticed that the output still contains germline indels. So perhaps the --NormalSamples hasn't been fully implemented yet and I'll still need to post-process results to get somatic indels from Pindel?

Thanks for any suggestions.

Answer 1

[Kevin Blighe]

A late answer but I am currently using Pindel.

Current versions (now June 8th, 2019) do perform somatic indel calling.

Here is my config file:

BB40.bam        150 Tumor
BM_Control.bam  150 Normal

Here are my commands to loop through each chromosome (mouse; excluding sex chromosomes):

mkdir pindel/ ;

for chr in {1..19}; do
  mkdir pindel/chr"${chr}"/

  /Programs/pindel/pindel \
    --fasta /ReferenceMaterial/mm10/mm10.fasta \
    --config-file pindelConfigs/BM39.txt \
    --chromosome chr"${chr}" \
    --number_of_threads 2 \
    --min_num_matched_bases 30 \
    --report_inversions TRUE \
    --min_inversion_size 50 \
    --report_duplications TRUE \
    --report_long_insertions TRUE \
    --report_breakpoints TRUE \
    -o pindel/chr"${chr}"/ ;

  /Programs/pindel/pindel2vcf \
    --pindel_output_root pindel/chr"${chr}"/ \
    --reference /ReferenceMaterial/mm10/mm10.fasta \
    --reference_name mm10 \
    --reference_date 99999999 \
    --min_coverage 1 \
    --het_cutoff 0.01 \
    --hom_cutoff 0.99 \
    --vcf pindel/chr"${chr}"/chr"${chr}".vcf ;

  bgzip -f pindel/chr"${chr}"/chr"${chr}".vcf ;

  tabix -f -p vcf pindel/chr"${chr}"/chr"${chr}".vcf.gz ;

  /Programs/bcftools-1.9/bcftools view -Ov --exclude-uncalled --min-ac=1 pindel/chr"${chr}"/chr"${chr}".vcf.gz > pindel/chr"${chr}"/chr"${chr}".filt.vcf ;
done

The resulting VCFs contain the calls for the tumor and normal sample, and indicate presence or absence in both.

Kevin

[steve]

thanks for this, exactly what I was looking for

[always_learning]

What if I have multiple pairs of data? Do I need to generate separate config file for each pair?

Regards, Najeeb

[Kevin Blighe]

Do I need to generate separate config file for each pair?

If you want separate somatic calls, I believe 'yes'.

I am not 100% confident, though, and pindel functionality changes depending on the version that you are using. Please review all command line options for the version that you are using, and [if possible], perform some tests with a few BAM files.