CREATIVE MIND, CREATIVE WORKS.

Dedicated to
High
Performance
Computing
In Quantitative Genetics

About us

We breath data

Quantitative genetics applied to animal and plant breeding has already been BIG DATA when neither word “BIG DATA” was invented nor the computers were suitable for it.

Still to today, the data we move and the models we write are posing a serious challenge even for large scale computers.

From genes to random numbers

We understand how biology can be described in the language of mathematics, and how genes are related to random numbers.

We are experts in translating quantitative genetics ideas into linear models and performing computer code.

Our solutions perform

Whether it is 50 million pedigree individuals, 5 million genotypes, a billion equations, or sophisticated models – whatever your challenge is, we have already a performing software solution. If not, we’ll make one.

We are naturally multilingual

We speak Bash, R, Julia, Python, C, C++, Fortran77, and Fortran08.

Whatever language your numerical problem speaks we will understand it.

Our Software

software we license to clients

APEX – Linear Mixed Model Software which performs

Endless functionalities

Our SINGLE-EXECUTABLE software encompasses all functionalities for a successful linear mixed model analysis:

  • Preconditioned gradient solver
  • Direct solver
  • Parallel chain Gibb sampler
  • Parallel chain Threshold Gibbs sampler
  • Exact reliability calculation
  • Deterministic reliability approximation
  • MCMC reliability approximation
  • Inverse matrix sampling
  • SUPER-AI-REML
  • single step EBV decomposition
  • Multi-trait multi-effect single-step de-regression

Since the equation system is never built, we happily deal with billions of equations.

All models are support

Our software supports all models currently used in genetic evaluation, including all models for genomic marker: ssGBLUP, ssAPYBLUP, ssGTBLUP, ssSNPBLUP.

Our ssSNPBLUP implementation performs in models with several million genotyped individuals and more than 1 billion equations, all WITHOUT terabytes of RAM or SSD swap space. It delivers multi-trait prediction equations and

  • breeding values,
  • direct genomic values,
  • residual poly-genetic values, and
  • imputation error values

for ALL pedigree individuals in on go. Headaches over “who is in the core” or marker pre-selection belong to the past.

Stable reliabilities

Quantitative geneticists may like random numbers …… anybody else doesn’t.

Deterministic reliability approximations for large multi-trait models including single step guarantee traceable changes, a must when serving members or clients.

Our single step reliability approximation:

  • can handle millions of genotypes in pedigrees of 10th of millions of individuals
  • allows for all genetic evaluation models including random-regression test-day
  • can deploy GBLUP or SNPBLUP on the fly
  • automatically adjust for genomic and poly-genetic weights

Multiple interfaces

Our software features multiple interfaces including an instruction file interface in YAML format and a dynamic library callable from R.

In addition our software can be used as a command line tool for adhoc task like pedigree analysis, NRM block extraction, GRM constructions, GRM block extractions, genotype duplication search and many more.

Modern programming

Our core programming language is modern C/C++, the de-facto standard for high performance computing. It allows us to write extremely fast code but is also flexible to reflect the real world problem we are modelling.

Thanks to C/C++ being widely used we can compile and test our code using many different compilers, and a vibrant community assures that the language is constantly evolving.

Deploy anywhere

We tailor interfaces and executable to the client’s needs.

Whether it is Linux, Mac, Windows or Android, Intel or AMD chips, local servers or the the cloud, specific interface layouts, specific data formats, or zero-interface executable.

Nothing is impossible.

Speedy extension

C++ features like Object Orientation and Template Meta Programming, combined with our flexible linear model frame allow for an almost instantaneous implementation of new features.

Whatever is necessary, we’ll make it happen in no time.

Ongoing support

Spurious results, long run time, abnormal resource usage, non-convergence, abnormal solutions, un-explainable changes in reliabilities.

Whatever the problem is we’ll investigate.

Ongoing development

Whether it is newly developed high performance computing libraries, graphical computing devices with teraflops of computing speed, advances in OpenMPI or OpenMP, optimization of our code base, or optimized reformulations of algorithms.

We constantly evolve our system to provide the fastest solutions to our clients.

Our Service

service we provide to clients

Software development

New algorithms often require new source code.

We support clients in expanding their existing code base to cover latest developments in science and technology with special focus on BIG DATA in quantitative genetics

Software review

The ever increasing volume of data will expose computational bottle necks in existing genetic evaluation pipelines.

We support clients in reviewing a non-performant code base and support the management with recommendations for operational and strategic decisions.

Software optimization

Generational change often means that the FORTRAN-ners retire and the Python-ians take over, rendering an entire code base unmaintained.

Our extensive knowledge in C, C++, and FORTRAN, combined with an in-depth understanding of quantitative genetics algorithms allows us to maintain and increase the performance of an existing code base in the most cost-effective way.

Pipelines

We design, develop, deploy, and maintain database-to-database pipelines for genetic evaluation and imputation.

Whether it’s the programming language, the interface design or diagnostics, everything will be tailored according to our clients’ needs.

Databases

Modern genetic evaluation is unthinkable without capable databases handling huge volumes of data. This even more applies when it comes to storing and retrieving genomic information.

We design, develop, deploy and maintain database systems for genetic evaluation in plants and animals with special focus on genomic data.

Algorithms

Ideas and algorithms sometimes need decades to be implemented in commercially viable software solutions. And sometimes algorithms for pressing practical problems don’t even exist.

We are specialized in development and programming of algorithms related to problems in quantitative genetics and genetic evaluation of plants and animals.

Technical backup

Changes in evaluation results can be hard to understand, consequences of new evaluation models difficult to fathom, and implications of hard- and software advances are sometimes not on the radar.

We support clients in situations which require an in-depth knowledge of genetics, mathematics, statistics and computer science.

Genetic evaluation

Sometimes all what’s needed is the results.

We support clients by running the entire genetic evaluation from scratch including finding the right models and parameters, organizing the computational resources and efficiently running the evaluation engine.

Research

From novel ideas to grant applications, publishable results and commercially viable implementations.

Our contract research service covers the entire scientific value chain.

We are specialized in making the impossible possible when timelines and budgets are tight.

Our Partners

Companies we support

Our Story

The foundations of our expertise

GHPC CONSULSTING AND SERVICES PTY. LTD. was founded in 2021 by Dr. Vinzent Boerner.

Dr. Boerner holds a BSc in Agriculture, a MSc in Animals Science, and PhD in Quantitative Genetics.

From 2011 till 2021, Dr. Boerner was working as a Researcher and Associate Professor at the Animal Genetics and Breeding Unit(AGBU), University of New England(UNE), Armidale, Australia, where he was the sole designer and developer of the linear model analysis software backing the Australian Beef and Sheep genetic evaluation. He also authored numerous scientific publications and invented two algorithms for marker based parentage verification and breed composition estimation.

From 2019 till 2023 Dr. Boerner also held an Associate Professor position at the Center for Quantitative Genetics and Genomics, Aarhus University, Denmark where he was the main developer of the Linear Models Toolbox.