hmcnc - Hidden Markov Copy Number Caller

Tool for calling CNVs based on read depth and SNV signatures

---

---

Requirements

• gxx
• meson
• ninja
• boost
• htslib
• zlib

install

• clone the repo
• build the repo with meson and ninja

cd hmcnc ; mkdir build ; cd build ; meson .. ; ninja

usage

Hidden Markov Copy Number Caller

Usage: ./src/hmcnc [OPTIONS] reference

Positionals:
  reference FILE REQUIRED     Read reference from this FASTA file.

Options:
  -h,--help                   Print this help message and exit

Input:
  -a FILE                     Read alignments from this BAM file and calculate depth on the fly.
  -b FILE                     Read depth bed from this file (skip calculation of depth).
  -s FILE                     Read SNVs from this file (when not estimating from a BAM).
  -p FILE                     Read parameter file (do not train with Baum-Welch).
  -l FILE                     Read clipping signature file (when not estimating from a BAM).

Depth Calculation:
  -e FLOAT                    Value of log-epsilon. [-800]
  -m TEXT                     Coverage model to use: Poisson (pois), or negative binomial (nb). [nb]
  -t INT                      Number of threads. [4]
  -c TEXT                     Use this contig to estimate coverage. By default, longest contig.

Output:
  -o FILE                     Output vcf to this file. Write to stdout if not provided.
  --sample TEXT               Sample name in the vcf ['sample']
  -M                          Merge consecutive bins with the same copy number.
  -C TEXT                     Only run hmm on this chrom.
  -B FILE                     Write coverage bed to this file.
  -P FILE                     Write trained parameter file.
  -S FILE                     Write SNVs to this file.
  -L FILE                     Stores the number of reads with clipping > 500 bases in each bin.

example

hmcnc human_GRCh38_no_alt_analysis_set.fasta -a HG002.GRCh38.deepvariant.haplotagged.bam -t 20 -o out.vcf