Why Systematic Noise Matters

Why Systematic Noise Matters

In somatic variant calling, sequencing and library preparation can introduce recurrent artifacts that appear across all samples. These artifacts can be mistaken for true somatic variants, leading to false-positive calls. This is especially problematic in tumor-only mode, where there is no matched normal sample to filter germline and artifact variants.

The systematic noise baseline builder processes a set of normal samples (without somatic variants) to identify these recurrent artifacts. The resulting noise file is then used during variant calling to filter out these artifacts and improve the specificity of your results.

IMPORTANT. Systematic Noise file creation is a one time effort per sample type + library prep + panel + sequencer combination.

Best Practices for Building the Noise Baseline

• For robust noise estimation use at least 30 normal samples.

• Match the library prep kit, target panel BED file, and sequencer platform of the normal samples to your tumor samples.

• Match the sample type (FFPE, fresh frozen, blood) and quality (DNA Integrity Number score range) to your expected tumor samples.

• Sequence at the same depth that the tumor samples are expected to be sequenced.

Required Inputs

• Input reads: FASTQ, BAM, CRAM, or ORA files (ORA supported on ICA only)

• Target panel BED file

IMPORTANT. The reference genome for your BED file, systematic noise analysis, and variant calling must be exactly the same for optimal results. The system will not raise an error if different reference genomes are used, but the results will be far from ideal.

Choose Your Option

Pick the path that matches your situation:

• Option A: From Run Planner — when you are planning a new sequencing run.

• Option B: From BSSH or ICA — when you already have sequencing data.

• Option C: Pre-Built Reference Files — fallback for initial evaluation only; not recommended for production use.