DISCOVERING PATHOGENIC VARIANTS

A human has about 22,000 genes, each of which may have up to 10,000 variants. Think about discovering a needle in the proverbial haystack! It is no wonder that many users of genomic data, including highly trained medical professionals, become easily overwhelmed by the abundance of data.

American College of Medical Genetics (ACMG) classifies all variants into 5 categories:

1. Benign
2. Likely benign
3. Likely pathogenic
4. Pathogenic
5. Uncertain significance or VUS

MoodNote Clinical Genomics-trained staff use sophisticated bioinformatic tools and complex algorithms to discover pathogenic or likely pathogenic variants responsible for disease, and deliver this information in an easy to comprehend report.  

In order to identify deleterious variants, we rely on four sets of criteria

1. Clinical significance

Listing of a variant in ClinVar database. If a variant is listed as “pathogenic” or “likely pathogenic” in ClinVar, it is because someone already studied this variant and found that it is indeed associated with a disease or symptom. If a variant listed as “benign” in ClinVar, there is a good chance that this variant is not linked with a particular disease or symptom.

2. Variant frequency in the population

How many people in the general population have a particular variant? Obviously, if most people have that variant, it is unlikely to be associated with a disease since most people in the population are not ill. If a variant is found in less than 5% of the population, there is a good chance that it may be damaging. In other words, the more rare a variant is, the more likely it is associated with a disease.

3. Computational (In silico)  predictions

In silico predictors are computer programs that calculate a potential impact of a variant on the protein structure and function. Our bioinformatics analysis factors in calculations and models made by multiple predictors. They include MutationTaster, SIFT, PROVEAN, metaRNN, CADD and others. Our algorithms take into account results of all predictors weigh them with other findings to issue a "verdict" based on ACMG rules.  

4. Conservation

If a variant appears in an evolutionary conserved part of the gene, it is more likely to be deleterious. There are several methods to calculate conservation scores. Our algorithms use several conservation metrics including GERP and phastCons. High scores indicate that a variant is more likely to be deleterious.

Report Generation

One critical aspect of our variant discovery process is a thorough literature search. We use sophisticated tools such as UMLS Metathesaurus and other databases to determine whether a gene with a pathogenic variant is indeed associated with the patient's symptoms or condition. This is a challenging task because there are thousands of sources and publications to review. Often, the patient's condition is very rare, with only a few small studies available. Our advanced algorithms automate this process, requiring only minimal human intervention for the final report.

Technical Aspects of Pathogenic variant Discovery

For those interested in the technical aspects of our pathogenic variant discovery process, please see the graphic below depicting a standard workflow of our bioinformatics analysis pipeline.