<?xml version="1.0"?>

<tool id="hgvs_add_coverage" name="Add poor/missing coverage info">
  <description>to an annotated HGVS table</description>
  <version_string>hgvs_add_missing_sample_coverage -v</version_string>
  <command interpreter="perl">hgvs_add_missing_sample_coverage $input_hgvs_annotated_table $out_hgvs_annotated_table $sample_name $sample_bam $max_homo_fn_coverage $max_het_fn_coverage $min_mapq $min_baq</command>
  <inputs>
    <param format="achri_annotated_snp_table" name="input_hgvs_annotated_table" type="data" label="Annotated variant calls table with HGVS syntax"/>
    <param name="sample_name" type="text" label="Name of sample to check for read coverage"/>
    <param name="sample_bam" type="data" format="bam" label="Mapped reads to be checked (BAM format)" help="For each variant in the input table that does NOT include the specified sample name in the Sources column, this BAM file will be checked for concordant reads"/>
    <param name="max_homo_fn_coverage" label="Minimum read coverage to exclude false negative homozygous variants" type="integer" min="1" value="3" />
    <param name="max_het_fn_coverage" label="Minimum read coverage to exclude false negative heterozygous variants" type="integer" min="1" value="20" />
    <param name="min_mapq" label="Minimum mapping quality score to consider a read" type="integer" min="1" value="20" help="(10=p-value 0.1, 20=p-value 0.01, etc.)"/>
    <param name="min_baq" label="Minimum base quality score to consider a read" type="integer" min="1" value="20" help="(10=p-value 0.1, 20=p-value 0.01, etc.)"/>
  </inputs>
  <outputs>
    <data format="achri_annotated_snp_table" name="out_hgvs_annotated_table" type="data" label="Variant table with poor/missing coverage data for ${sample_name}"/>
  </outputs>

  <tests>
  </tests>

  <help>
  Once this tools is run, lack of the given sample name in a table row means the sample almost definitely does not contain the variant that row reports.
  This tool adds a sample to the Sources column of the variant table if it is not already there, but the given BAM file either:

1. has at least 1/max_het_fn_coverage read proportion supporting the variant (in which case the sample name is added with a "+"), or

2. has fewer than max_homo_fn_coverage or max_het_fn_coverage reads mapped in total, and those reads neither conflict with nor support the variant called (in which case the sample name is added with a "~")
  </help>

</tool>