Variants in which the reference allele has a very low allele fraction are marked for filtering in the cancer model

Describe the bug Variants in which the reference allele has a very low allele fraction are marked for filtering by the AF filter

Version

$ octopus --version
octopus version 0.7.2 (728bdb81)
Target: x86_64 Linux 3.10.0-1127.18.2.el7.x86_64
SIMD extension: AVX2
Compiler: GNU 9.3.0
Boost: 1_75

Command Command line to install octopus:

$  octopus/scripts/install.py --dependencies --forests

Command line to run octopus:

'octopus \
                                -C cancer \
                                --allow-marked-duplicates \
                                --allow-octopus-duplicates \
                                --min-expected-somatic-frequency 0.05 \
                                --min-credible-somatic-frequency 0.01 \
                                --max-somatic-haplotypes 2 \
                                --filter-expression "QUAL < 10 | MQ < 10 | MP < 10 | AD < 1 | AF < 0.01 | AFB > 0.95 | SB > 0.98 | BQ < 15 | DP < 1 | ADP < 1" \
                                --somatic-filter-expression "QUAL < 2 | GQ < 20 | MQ < 30 | SMQ < 40 | SB > 0.90 | SD > 0.90 | BQ < 20 | DP < 3 | ADP < 1 | MF > 0.2 | NC > 1 | FRF > 0.5 | AD < 1 | AF < 0.0001" \
                                --annotations SB SD AF \
                                --threads \
                                -w temp/ \
                                -I {input.tumour_bams} \
                                --regions-file {input.interval_file} \
                                --output {output.tumour_vcf} \
                                -R {input.reference_genome}'

Additional context Add any other context about the problem here, e.g.

Reference: GRCh37_g1kp2 (i.e. version of GRCh37 developed/released for use in the 1000 genomes project)
Sequencing data: Illumina amplicon sequencing data. 150bp paired-end

Hello,

I am not sure if this is intended behaviour or not, but octopus seems to be labelling variants in which the reference allele has a very low allele fraction for filtering by the AF filter. For example:

Apologies in advance if I have misunderstood the intended behaviour of the tool. I have not tested this on other calling models.

Thanks!

luntergroup / octopus

Variants in which the reference allele has a very low allele fraction are marked for filtering in the cancer model #173