precisionlife MARKERS 2018-05-02T21:38:03+00:00

2018 – PRECISIONLIFE MARKERS BRCA2 PAPER NOW AVAILABLE
CLICK HERE FOR MORE INFORMATION

multi-modal, polygenic biomarker studies in hours or days, not months
  • rapidly identifying biomarker networks with up to 50 genomic, phenotypic and clinical factors acting in combination

  • comprehensive, reproducible and interpretable results in hours or days on a single 4 GPU machine

  • integrated deep semantic annotation for validation and biological interpretation, aiding discovery of new, more specific and useful biomarker networks

  • enabling detailed patient stratification in clinical trials design, healthcare analytics & clinical decision support

  • affordable multi-modal, multi-dimensional, hyper-combinatorial analysis of disease population scale genomic study data

read more in the IBM Technical Briefing…

precisionlife MARKERS is a powerful new analytical approach to multi-modal biomarker discovery and validation studies.

precisionlife MARKERS (previously known as Synomics Studio) enables large scale datasets including combinations of:

  • genomic
  • phenotypic, and
  • clinical features

to be fully analyzed for multi-feature associations (with up to 30 features) that meet stringent user-defined criteria for p-value, penetrance and SNP network size.

precisionlife MARKERS uses a massively parallelized algorithm, tuned for multiple GPU compute devices, to give results in minutes, hours or days, rather than weeks or months.

precisionlife MARKERS makes it easy for genomics and precision medicine researchers to quickly identify, validate and understand the metabolic relevance of disease related features in novel biomarker networks by combining multi-factor association analysis with integrated validation and biological annotation and interpretation features. These include a full bioinformatics knowledge graph containing SNP, gene & pathway annotations and systems pharmacology models.

precisionlife MARKERS annotation engine precisionlife MARKERS can be used for:

  • more accurate patient stratification
  • personalized disease risk scoring
  • assessment of likely therapy response / adverse drug reactions

Example Study:

A study was performed on a population of around 15,000 people, all of whom are BRCA1/2 mutants and therefore have higher risk of developing breast, ovarian, and other cancer risks (such as prostate in males) due to abnormal DNA repair and recombination functionality.

The cohort is split into those ‘affected’ individuals who have or have had ‘early’ onset breast cancer (<40 years) and ‘unaffected’ who have not developed breast cancer by age 55. For BRCA2 mutants (around 1,500 affected and 6500 non-affected people) the analysis was run both in the usual fashion comparing affected ‘cases’ to non-affected ‘controls’ (to find disease risk factors), as well as non-affected to affected ‘controls’, to find disease protective factors.

This analysis has identified 3 almost completely non-overlapping sub-cohorts within the affected groups, indicating potentially different molecular etiologies for the disease. The study has identified a number of (novel) statistically significant protective combinations of SNPs which reduce a person’s risk of developing breast cancer significantly, even when they have a BRCA2 mutation. It also found consistently co-associating sub-clusters of SNPs which are associated with specific pathways that co-occur, and a range of genes that are newly implicated in breast cancer in specific combinations with other more well known variant alleles.

The study results showed that even for people with disease risk mutants, there are no biomarker networks containing less than 6 SNPs in combination that are represented only in the affected population and not in the unaffected. This helps explain the lack of performance of traditional GWAS (which can combine only a maximum of a couple of SNPs at a time) in complex diseases, as networks with combinations that appear in the control set may well be random observations.

The most populous biomarker networks identified contain up to 17 SNPs acting in combination – a complexity of analysis and level of insight into the disease and personal risk that only precisionlife MARKERS can achieve.

precisionlife MARKERS (previously known as Synomics Studio) runs on various architectures including GPU enabled systems such as IBM Minsky, and is featured in the IBM Reference Architecture for High Performance Analytics in Healthcare and Life Science and in this IBM Technical Briefing

multi-modal, polygenic biomarker studies in hours or days, not months
  • rapidly identifying biomarker networks with up to 50 genomic, phenotypic and clinical factors acting in combination

  • comprehensive, reproducible and interpretable results in hours or days on a single 4 GPU machine

  • integrated deep semantic annotation for validation and biological interpretation, aiding discovery of new, more specific and useful biomarker networks

  • enabling detailed patient stratification in clinical trials design, healthcare analytics & clinical decision support

  • affordable multi-modal, multi-dimensional, hyper-combinatorial analysis of disease population scale genomic study data

read more in the IBM Technical Briefing…

precisionlife MARKERS is a powerful new analytical approach to multi-modal biomarker discovery and validation studies.

precisionlife MARKERS (previously known as Synomics Studio) enables large scale datasets including combinations of:

  • genomic
  • phenotypic, and
  • clinical features

to be fully analyzed for multi-feature associations (with up to 30 features) that meet stringent user-defined criteria for p-value, penetrance and SNP network size.

precisionlife MARKERS uses a massively parallelized algorithm, tuned for multiple GPU compute devices, to give results in minutes, hours or days, rather than weeks or months.

precisionlife MARKERS makes it easy for genomics and precision medicine researchers to quickly identify, validate and understand the metabolic relevance of disease related features in novel biomarker networks by combining multi-factor association analysis with integrated validation and biological annotation and interpretation features. These include a full bioinformatics knowledge graph containing SNP, gene & pathway annotations and systems pharmacology models.

precisionlife MARKERS can be used for:

  • more accurate patient stratification
  • personalized disease risk scoring
  • assessment of likely therapy response / adverse drug reactions

Example Study:

A study was performed on a population of around 15,000 people, all of whom are BRCA1/2 mutants and therefore have higher risk of developing breast, ovarian, and other cancer risks (such as prostate in males) due to abnormal DNA repair and recombination functionality.

The cohort is split into those ‘affected’ individuals who have or have had ‘early’ onset breast cancer (<40 years) and ‘unaffected’ who have not developed breast cancer by age 55. For BRCA2 mutants (around 1,500 affected and 6500 non-affected people) the analysis was run both in the usual fashion comparing affected ‘cases’ to non-affected ‘controls’ (to find disease risk factors), as well as non-affected to affected ‘controls’, to find disease protective factors.

This analysis has identified 3 almost completely non-overlapping sub-cohorts within the affected groups, indicating potentially different molecular etiologies for the disease. The study has identified a number of (novel) statistically significant protective combinations of SNPs which reduce a person’s risk of developing breast cancer significantly, even when they have a BRCA2 mutation. It also found consistently co-associating sub-clusters of SNPs which are associated with specific pathways that co-occur, and a range of genes that are newly implicated in breast cancer in specific combinations with other more well known variant alleles.

The study results showed that even for people with disease risk mutants, there are no biomarker networks containing less than 6 SNPs in combination that are represented only in the affected population and not in the unaffected. This helps explain the lack of performance of traditional GWAS (which can combine only a maximum of a couple of SNPs at a time) in complex diseases, as networks with combinations that appear in the control set may well be random observations.

The most populous biomarker networks identified contain up to 17 SNPs acting in combination – a complexity of analysis and level of insight into the disease and personal risk that only precisionlife MARKERS can achieve.

precisionlife MARKERS (previously known as Synomics Studio) runs on various architectures including GPU enabled systems such as IBM Minsky, and is featured in the IBM Reference Architecture for High Performance Analytics in Healthcare and Life Science and in this IBM Technical Briefing