Cadd Score Calculator

Use this interactive calculator to interpret a CADD PHRED score, estimate its deleteriousness percentile, and place a genomic variant into a practical evidence tier. CADD, or Combined Annotation Dependent Depletion, is widely used in variant prioritization because it compresses many genomic annotations into a single score that helps researchers and clinicians rank potentially harmful variants.

What the CADD score calculator measures

The CADD score calculator on this page is designed to help users interpret a CADD PHRED score in a fast, consistent, and clinically informed way. CADD stands for Combined Annotation Dependent Depletion. It is a framework that integrates a very large number of annotations about genomic position, conservation, protein effect, regulatory data, and other features into a single estimate of how deleterious a variant may be. In practice, many genomic analysts use CADD as one of several ranking tools when reviewing candidate variants from exome or genome sequencing.

The most commonly discussed value is the scaled PHRED-like CADD score. This transformed score is easier to interpret than the raw score because it gives a relative ranking against all possible single nucleotide variants in the human genome. A higher CADD PHRED score means a variant is predicted to be more deleterious relative to the genomic background. For example, a score of 10 indicates the variant is among the top 10% most deleterious substitutions, 20 indicates the top 1%, 30 indicates the top 0.1%, and 40 indicates the top 0.01%.

That ranking logic is exactly why a calculator like this is useful. The numerical score alone can be misread if a user does not remember what the PHRED thresholds mean. By turning the raw number into a percentile interpretation, severity tier, and variant-specific note, the calculator reduces ambiguity and supports more disciplined review. It does not replace expert judgment, disease-specific rules, segregation analysis, population frequency review, or functional validation. Instead, it acts as an interpretation aid in a larger evidence framework.

How to interpret CADD PHRED thresholds

The scaled PHRED score is intentionally ordinal. It works best as a ranking system, not as standalone proof of pathogenicity. A common mistake is to treat a high CADD score as a diagnosis. That is not how the metric should be used. A high score should prompt closer review, while a low score may reduce priority in broad filtering pipelines. The score should always be integrated with conservation, allele frequency, disease mechanism, inheritance pattern, transcript relevance, and phenotype correlation.

CADD PHRED score	Approximate ranking among possible variants	Interpretation	Typical analytic use
0 to 9.9	Below top 10%	Lower predicted deleteriousness	Often lower priority unless strong orthogonal evidence exists
10 to 19.9	Top 10% to top 1%	Moderately elevated	Useful for triage in broad candidate lists
20 to 29.9	Top 1% to top 0.1%	High predicted deleteriousness	Frequently retained for deeper review and clinical correlation
30 to 39.9	Top 0.1% to top 0.01%	Very high predicted deleteriousness	Strong prioritization signal, especially in constrained genes
40 and above	Top 0.01% or rarer	Extreme predicted deleteriousness	Rarely ignored, but still requires complete evidence review

Why the PHRED-style scale matters

Many genomic tools produce scores on scales that are hard to compare. CADD is attractive because the scaled PHRED score compresses complex model output into an intuitive ranking system. A score of 20 is not merely twice as concerning as a 10. Rather, it reflects a substantially more selective ranking among possible variants. This distinction is important in real workflows because variant reviewers are often trying to narrow tens of thousands of candidates to a much smaller, phenotype-relevant shortlist.

Typical rules of thumb in practice

• Scores below 10 are usually lower priority unless the variant has powerful corroborating evidence.
• Scores of 15 or higher often draw attention in rare disease pipelines.
• Scores of 20 or higher are commonly treated as a meaningful deleteriousness threshold.
• Scores of 30 or higher are often considered especially notable, but not automatically pathogenic.
• Very high scores in genes with strong disease validity may deserve rapid review if phenotype and inheritance also fit.

How this calculator computes the result

This calculator uses the standard interpretation of the scaled CADD PHRED score. The primary mathematical step is mapping the entered PHRED score to an approximate percentile tier:

1. If the score is below 10, the variant is interpreted as outside the top 10% most deleterious substitutions.
2. If the score is from 10 to 19.9, it falls within the top 10% to top 1% range.
3. If the score is from 20 to 29.9, it falls within the top 1% to top 0.1% range.
4. If the score is from 30 to 39.9, it falls within the top 0.1% to top 0.01% range.
5. If the score is 40 or higher, it is interpreted as top 0.01% or more extreme.

After that, the calculator adds context from the selected variant type and gene constraint setting. These selections do not alter the CADD score itself, because CADD is already computed upstream. Instead, they generate a more useful interpretation narrative. For example, a high CADD score attached to a nonsense or frameshift variant in a highly constrained gene is generally more compelling than the same score attached to a synonymous variant in a gene with low known constraint.

Important: CADD should not be used in isolation to classify pathogenicity. It is a prioritization metric, not a diagnostic verdict. Final variant interpretation should incorporate population frequency, ClinVar or curated databases, gene-disease validity, segregation data, inheritance pattern, transcript selection, functional studies, and phenotype match.

Where CADD fits in a real variant interpretation workflow

In practical genomics, CADD usually appears after primary variant calling and annotation. Once variants are aligned to a reference genome and basic annotation is complete, analysts begin filtering by quality, allele frequency, expected inheritance model, and variant consequence. This can still leave a large list. CADD helps rank that list. Instead of reading every candidate equally, a team can focus first on variants with stronger computational evidence of deleteriousness.

Consider a rare disease exome study. A trio might produce thousands of rare coding or splice-region variants before filtering. If only a handful fit the phenotype and inheritance model, CADD may be a helpful tie-breaker. A de novo missense variant with a CADD score of 28 in a constrained neurodevelopmental gene usually deserves more attention than a synonymous variant with a score of 6 in a gene with weak disease evidence. The key is not that CADD makes the decision by itself, but that it helps rank plausibility efficiently.

In cancer genomics, the logic is similar but not identical. Somatic interpretation often depends on tumor context, hotspot status, clonality, pathway relevance, and treatment implications. CADD can still be informative, but it is not a substitute for tumor-specific interpretation frameworks. Likewise, in population studies, CADD may support burden testing or variant weighting, yet careful study design and replication remain essential.

Comparison table: CADD versus other common prioritization signals

Signal	What it measures	Typical strength	Main limitation
CADD PHRED	Integrated deleteriousness ranking across many annotations	Strong for broad prioritization and ranking	Not disease-specific and not sufficient for final classification
Population frequency	How common a variant is in reference populations	Very strong for filtering implausibly common variants	Rare does not equal pathogenic
Consequence class	Predicted molecular impact such as missense or stop gained	Strong when aligned with known disease mechanism	Consequence alone may overestimate effect
Segregation data	Whether the variant tracks with disease in a family	High evidentiary value in many settings	Often unavailable in small families or sporadic cases
Functional assay	Experimental evidence of altered biological effect	Potentially decisive if assay is validated	Assay relevance and reproducibility can vary

Real threshold statistics that matter when using a CADD score calculator

One of the most useful aspects of CADD is that several benchmark thresholds correspond to intuitive rarity levels among possible variants. These are not arbitrary clinical bins. They are tied to the PHRED-like transformation and make the score interpretable at a glance.

• A CADD PHRED score of 10 corresponds to roughly the top 10% most deleterious substitutions.
• A score of 20 corresponds to roughly the top 1% most deleterious substitutions.
• A score of 30 corresponds to roughly the top 0.1% most deleterious substitutions.
• A score of 40 corresponds to roughly the top 0.01% most deleterious substitutions.

These thresholds are why many pipelines use 15, 20, or 25 as practical cut points depending on desired sensitivity and specificity. A highly exploratory research workflow may retain all variants above 15. A stricter clinical review pipeline may focus on 20 and above after phenotype and frequency filters are applied. There is no universal threshold that works for every disease area, which is exactly why a transparent calculator is useful. It translates the score without pretending that one cut point solves every interpretation problem.

Best practices for using this tool responsibly

1. Start with high-quality variant annotation

CADD interpretation is only as reliable as the annotation workflow that generated the score. Make sure the transcript model, reference build, and consequence term are correct. A build mismatch can create confusion, especially when comparing GRCh37 and GRCh38 resources.

2. Pair the score with phenotype and inheritance

A very high CADD score in the wrong gene or the wrong inheritance context is less useful than a moderately high score in a gene that fits the disease mechanism. Variant review should always begin with the clinical or biological question, not with the computational score alone.

3. Check population databases

Even variants with elevated CADD scores can appear in population reference datasets at frequencies that are inconsistent with severe Mendelian disease. If a variant is too common for the phenotype under study, that usually overrides a high computational deleteriousness estimate.

4. Use orthogonal evidence

Conservation metrics, splice prediction tools, missense meta-predictors, gene constraint scores, clinical databases, and functional data all add context. The strongest interpretations come from concordant evidence across multiple independent sources.

5. Document the threshold you applied

If you use a CADD score cutoff in a pipeline, state it clearly and explain why. This improves reproducibility and makes it easier for other reviewers to understand your filtering strategy.

Authoritative resources for deeper reading

If you want to validate interpretation principles or learn more about variant classification standards, review these authoritative sources:

• National Center for Biotechnology Information (NCBI)
• National Human Genome Research Institute (genome.gov)
• MedlinePlus Genetics from the U.S. National Library of Medicine

Final takeaway

A CADD score calculator is most valuable when it converts a raw PHRED score into a disciplined interpretation. Higher scores indicate stronger predicted deleteriousness, but the score is still one piece of a broader evidence puzzle. Use the calculator to understand percentile rank, identify when a variant crosses meaningful thresholds, and create a concise explanation for reports or review meetings. Then move beyond the number: verify the transcript, check frequency data, align the finding with the phenotype, and integrate disease-specific evidence. That combined approach is where responsible genomic interpretation begins.