Disease Studies

Proving the Power of Combinatorial Analytics

Validation of COVID-19 Insights

70% of novel gene targets identified to be associated with serious disease and hospitalization with combinatorial analytics by PrecisionLife in May 2020 have subsequently been independently validated in the global COVID-19 scientific literature.

Proving the Power of Combinatorial Analytics

Proving the Power of Combinatorial Analytics: Validation of COVID-19 Insights

Krystyna Taylor

by Krystyna Taylor, Senior Portfolio Manager, PrecisionLife


At a time when the genetic predisposition to severe COVID-19 was still a mystery, PrecisionLife’s groundbreaking 2020 study on risk factors in severe COVID-19 patients[i] identified 68 novel genes that were associated with serious disease and hospitalization in the first small UK Biobank COVID-19 population. Since then, over 70% of these gene targets have subsequently been independently validated in the global COVID-19 scientific literature, ranging from massively powered multinational GWAS meta-analyses to phase 2 clinical trials.

In May 2020 SARS-CoV-2 was a new threat whose clinical implications for different patients were unclear.

There were no approved vaccines for COVID-19, the RECOVERY trial results had ended hopes of any clinical benefit of hydroxychloroquine, and the only genetic associations identified for severe COVID-19 risk were linked to the ABO gene and a narrow region on chromosome 3 [ii]. We didn’t understand the range of symptoms of the disease or why some patients got mild symptoms and others had life threatening disease.

At PrecisionLife we seized the opportunity to analyze the first small cohort of 779 severe COVID-19 patients reported in the UK Biobank. We wanted to shed light on which host genes were implicated with the most serious forms of the diseases, and why there were so many different symptoms.

COVID-19 analysis of genetic host response risk factors in severe COVID-19 patients

Finding the first genes linked to hospitalization risk

Our analysis, completed and written up two weeks later, demonstrated that 68 protein-coding genes had a significant association with hospitalization risk. For all but two there was no published evidence linking these genes to COVID-19, however PrecisionLife categorized them into 6 main pathways that could be associated with dysregulated host immune response to SARS-CoV-2:

  1. Viral replication
  2. Inflammation
  3. Endothelial cell dysfunction (associated with micro-coagulation)
  4. Lipid droplet biology (associated with reduced cell membrane integrity)
  5. Wnt/β-catenin signaling (associated with senescence and apoptosis)
  6. Alzheimer’s associated neurodegeneration (MAPT region)

This provided a range of plausible – if untested – hypotheses around the biological impact of variants in these genes on COVID-19 severity and symptoms.

Worldwide research validates combinatorial analytics

The global scientific efforts since we completed our initial analysis in June 2020 have been prodigious.

48 of the 68 genes identified in our initial genetic analysis using combinatorial analytics have now been independently published in articles in relation to SARS-CoV-2 replication or development of severe COVID-19.

The significance of some targets has been replicated in larger multinational genetic studies, whilst others have been evaluated in preclinical in vitro studies, or are targeted by drugs currently in phase II clinical trials for COVID-19.

This body of evidence provides retrospective validation for the genetic signals found that were entirely novel to the disease at the time of our initial COVID-19 work at PrecisionLife.

High resolution patient stratification of COVID-19 population with PrecisionLife's combinatorial analytics

Disease architecture showing high resolution patient stratification from PrecisionLife analysis of severe Covid-19 patients in May 2020 – colors represent patient subgroups, circles represent disease associated SNPs and lines represent co-associated SNPs


Examples of novel discoveries retrospectively validated:

1. Genes

After analyzing fewer than 1,000 cases, PrecisionLife was the first to report a significant association of 5 genes – KANSL, LINC02210, MAPT, SHT and SPPL2C – from the 17q21.31 locus in severe COVID-19 development.

These have subsequently been validated by the results from the COVID-19 Host Genetics Initiative in July 2021[iii], who performed three genome-wide association meta-analyses of 49,562 severe COVID-19 patients with over 2 million mild COVID-19 patients and demonstrated a significant association between this group of genes and hospitalization after SARS-CoV-2 infection.

2. Proteins & Pathways

There is also now significant preclinical evidence as to how many of the genetic variants identified by PrecisionLife impact the course of COVID-19 disease development.

For example, we hypothesized that variants in MLKL could lead to over-activation of the necroptotic inflammatory response, resulting in organ damage, and proposed that an MLKL inhibitor – necrosulfonamide – could reduce COVID-19-associated inflammation. There is now evidence that MLKL is highly upregulated in immune and epithelial cells in the lungs of severe COVID-19 patients and that necrosulfonamide inhibits the release of inflammatory cytokine IL-1β in response to SARS-CoV-2[iv].

3. Drugs

PrecisionLife’s genetic signals indicated the potential of 29 drugs and clinical candidates for use in treatment strategies to improve clinical outcomes in severe COVID-19 patients.

13 of these candidates have now been evaluated in COVID-19, ranging from targets in preclinical work investigating impact on SARS-CoV-2 replication and inflammatory cytokine secretion in vitro, to 14 different clinical trials (recruiting/active/completed) measuring the impact of these drugs on disease severity reduction in COVID-19 patients.

The results for many of these clinical trials have not yet been posted, although one that has, dutasteride – an inhibitor of the PrecisionLife COVID-19 gene target SRD5A1 – reduced SARS-CoV-2 viral shedding and inflammation in a randomized, double-blind, placebo-controlled interventional trial[v].

Combinatorial analytics finds deeper insights faster from smaller patient datasets

We performed our analysis using only genetic data from fewer than 1,000 COVID-19 patients in the UK Biobank at a time when standard analytical approaches such as GWAS failed to find useful signal from much larger populations.

Although the targets identified by PrecisionLife were almost entirely novel at the time of publication, over 70% of them have now demonstrated strong scientific evidence in SARS-CoV-2 infection or COVID-19 development. This shows the power of the combinatorial analytics approach to identify the drivers of complex disease biology in specific patient subgroups with implications for improving novel target discovery, precision drug development, clinical trials, and personalized healthcare.

Further information on our COVID-19 research, including links to our published studies, are available at precisionlife.com/covid19

Understand disease biology better

We find more signal in patient data than anyone else on the planet, generating deeper insights into disease biology that drive:

  • Novel drug discovery and precision repositioning to address unmet medical needs
  • Patient stratification biomarkers to identify and select responders for successful clinical trials
  • Better diagnostic and clinical decision support tools to predict, diagnose and treat disease

If you are innovating to improve patient outcomes in drug discovery, development or healthcare, PrecisionLife is your partner for driving precision medicine into chronic diseases.

Enter your details below to start a discussion that could improve millions of lives.

More News & Media

View All News & Media

Can AI Help Us Solve the Population Health Crisis?

PrecisionLife Co-founder and CEO, Steve Gardner, recently sat down with Inside Precision Medicine, the trade publication for molecular diagnostics…

Read more

Supporting Women and Girls in Science

We hope to help support and inspire both young investigators and established female leaders in the field, especially on…

Read more

Why drug discovery & development needs fixing for chronic diseases?

The blockbuster model for drug discovery and development in chronic diseases is inefficient and costly. We need a new…

Read more