IGS Around the Web

IGS Around the Web (7)

Tuesday, 14 June 2016 15:58

The IGS Implementation of BOLT

Written by

Bruce Golden
Theta Solutions

Over the last 50 years we have had evolution of the statistical methods used to calculate genetic predictions, EPDs, for livestock. What drove the evolution of these methods? Knowledge of statistical models? New methods? Data? Enabling computer technology? Golden states that he believes the drive for better models has been a desire to increase the accuracy of prediction.

Golden and Garrick had written grants to write genetic prediction software in the past. This avenue appears to have dried up, so they decided to start a company, Theta Solutions, in order to fund the development of genetic prediction. The latest genetic prediction runs contained 46,000 animals with genomic data.

Theta Solutions uses graphical processing units, originally built for video gaming, to have a high performance computer at a relatively low cost. The BOLT software focuses on custom turnkey analyses, once the system is set up all one needs to do is feed it data.

Using non-GPU computing, Golden can solve 51 million equations in 1649 seconds. The fastest GPU implementation took 78 seconds.

Why do we use a Bayesian sampler for solving mixed models?

  • No accuracy approximation bias 
  • Can get PE covariance
  • Can apply marker selection methods
  • Can include prior information


With traditional methods, it took 23 seconds per sample, with new implementation can do a sample in 2 seconds. (Gibbs sampling is kind of like turning a statistical crank over and over to solve very complex equations, each sample is one turn of the crank.) They also parallelized the sampling, further speeding up the process. This parallelized processing is like working cattle with 100s of chutes rather than a single cute.

There are three ways to combine genomics with traditional EPDs,

  • blending Genomic BLUP (combine pedigree prediction with genomic prediction, two separate analyses)
  • single-step Genomic BLUP (combine pedigree relationships and genomic relationships, one analysis)
  • hybrid model (single step with marker effects)


Single-step genomic models outperform traditional EPDs. But, the hybrid model outperforms both models, especially for unproven animals. The purpose of the hybrid model is to squeeze more information out of the data.

Currently looking at a data set with 6 million pedigree records, 4.8 million birth weight records, and 1.9 million post weaning gain records, 46,402 genotyped animals and used 44,414 SNP markers.

 

Read the full article here

Sunday, 17 April 2016 15:52

BOLT software brings more reliability to EPDs

Written by
The accuracy of genetic prediction took a big leap forward when genomically enhanced EPDs became available. Now, with new technology, prediction accuracy may get honed to an even finer edge.

Wes Ishmael | Aug 01, 2016

Less risk in genetic selection via higher-accuracy expected progeny differences (EPDs).

That’s the bottom line promise of BOLT, new software adopted by International Genetic Solutions (IGS) to conduct its National Cattle Evaluation (NCE).

IGS is the unique collaboration of 12 breed associations in the U.S. and Canada to conduct a common, multi-breed NCE with a combined database representing almost 17 million animals.

Many at the recent annual meeting of the Beef Improvement Federation heralded BOLT as a revolutionary step in NCE.

“It's a game-changer,” believes Wade Shafer, executive vice president of the American Simmental Association (ASA).

True accuracy

BOLT is an acronym for Biometric Open Language Tools. Bruce Golden and Dorian Garrick — noted animal breeders familiar to those in the seedstock business — developed the new software and license its use through their company, Theta Solutions LLC.

Increased prediction accuracy with BOLT comes through the software’s ability to directly incorporate genomic data into the EPD calculation.

“Until now, we’ve had to incorporate DNA information through a post-evaluation blending process that combines the independent genomic data and traditional EPD into one published EPD value,” says Larry Keenan, director of breed improvement for the Red Angus Association of America (RAAA). “BOLT gives us the capability to incorporate genomic information directly into EPDs.”

Even though systems before BOLT meant DNA data were blended post-evaluation, its inclusion in recent years significantly increased the EPD accuracy for young animals. “This absolutely will allow us to maximize use of information in the database for pedigree, phenotypic and genotypic data,” Keenan says.

“There is so much information, but because of the limitation of the models and technology that have been available to us previously, we couldn’t squeeze out all of the value from that information. We’ll be able to get more out of it with BOLT,” he says.

Another powerful feature is BOLT’s capacity to calculate actual (true) accuracy, says Shafer.

“Prior to BOLT, the calculation of accuracy in NCE has been limited to what has been dubbed ‘approximation methods,’ ” Shafer says. “These methods take an indirect approach to the calculation of accuracy and routinely result in estimates of accuracy that can be quite different than an animal's actual accuracy.”

Because BOLT can calculate accuracy directly, it is capable of producing true accuracy. “Prior to BOLT, it was thought that the calculation of true accuracy was not feasible on large databases due to the computational requirements being untenable,” Shafer says.

Keep in mind that EPDs are already the gold standard of genetic prediction. Given available technology, they’re as reliable as possible.

Increased precision

The massive size of the IGS database also lends itself to increased prediction accuracy.

“Accuracy is critical to making genetic progress,” Shafer emphasizes. “There are almost 17 million animals in the IGS database. The accuracy of prediction generally increases with the volume of data.”

Depending on your leanings, a multi-breed NCE also offers sturdier predictions than a single-breed analysis.

Think of it like this — a registered Red Angus bull has an EPD within that breed, built upon the pedigree, phenotypic data, and now, genotypic data — submitted to that breed’s database. But the bull may also sire or be a relative to lots of other cattle accounted for in other breed databases. Being able to account for more progeny and relatives across more of the cattle population makes for a more accurate snapshot of the bull's genetic potential.

ASA began working with Cornell University in 1995, funding research and development for multi-breed NCE. Unlike in the traditional model, where universities developed NCE models and performed NCE for breed association clients, ASA shared the development costs and ownership.

At the time, Simmental breeders — like those of other Continental breeds — were primarily focused on breeding up cattle to purebred levels.

“We had lots of halfblood, three-quarter-blood and seven-eighths-blood Simmental, but we were using a single-breed model for genetic evaluation,” Shafer explains. A single-breed model couldn’t fully account for breed differences or heterosis.

These days, of course, another advantage of multi-breed analysis is that hybrid and composite seedstock are commonplace.

What you can expect

“From what we have seen thus far, it appears BOLT has the capacity to improve the accuracy of prediction well beyond any currently existing technology,” Shafer says. “As is typically the case with new genetic evaluation technology, the biggest gains will be on younger, lower-accuracy animals — which, fortuitously, is the vast majority of animals being considered as breeding stock.”

When IGS begins conducting its NCE with BOLT, seedstock producers will likely see some reranking of bulls, which always causes discomfort. The EPD accuracies for some bulls will likely decline, too. Rather than an indication that the new method is less robust, declining accuracy with the initial BOLT NCE would mean that the previous accuracy, calculated via the approximation method, was too high.

Already, EPDs can be compared directly among IGS breeds for growth and carcass traits. Ultimately, plans call for making that possible with heifer pregnancy and stayability, too.

“BOLT is a software package that is very dynamic and flexible, with the ability to accommodate most any type of statistical model,” Shafer explains. “Its flexibility makes upgrading models or developing EPDs for new traits highly feasible.”

BOLT is lightning-fast, too, compared to current NCE models and technology. Running the current IGS NCE without BOLT takes a couple of days. Running the NCE on the same computer with BOLT only takes a few minutes.

“The remarkable increase in speed is due to the software being written in a manner that utilizes the hardware to its fullest capacity — an approach that takes a world-class understanding of the interaction between computer programming and hardware,” Shafer explains. “BOLT’s dramatic speed allows for computational feats that were previously considered untenable, such as the calculation of true accuracy on a massive database.”

IGS is in the process of readying its massive database for BOLT. There is no launch date, but indications are that it should be this year or the first part of next year.

 “Eventually, I believe all genetic evaluation providers, regardless of species, will go to it [BOLT] because of the power it brings to the table,” Keenan says. 

Read the full article here.http://www.beefmagazine.com/cattle-genetics/bolt-software-brings-more-reliability-epds

 

Alison Van Eenennaam | University of California, Davis

The application of genomics to improve the accuracy of EPDs is a rapidly developing field. There are ongoing improvements in genotyping and sequencing technologies, statistical methods to increase the correlation between genomic predictions and true genetic merit, and the computing systems to handle the large datasets associated with animal breeding. One thing still remains true in the genomic age and that is the need to collect accurate phenotypic records. It is essential to ensure performance data, pedigree, and DNA information are recorded and reported accurately. Genomic predictions will only be as reliable as the data upon which they are based.  Although it might seem like the genomics era could signal the end of performance recording, the opposite is true. Now more than ever, it is important that producers accurately report data, and ensure that animals which are genotyped are correctly identified so that their information can contribute towards improving the accuracy of the genomic predictions of the future.

Continue reading

Tuesday, 17 January 2017 11:20

Bob Weaber from KState and Larry Kuehn from US MARC

Written by

 

"Brown Bagger" electronic presentation a couple weeks ago. Bob's focus was on the changing landscape in the area of genetic evaluation and Larry talked about MARC's involvement in multi-breed and across breed comparisons. This is a highly recommended presentation for those interested in genetic evaluation.  Click here.

Tuesday, 17 January 2017 11:14

IGS: We Collaborate You Profit

Written by

BeefMagazine.com    |    Published  September 13, 2016
Chip Kemp, American Simmental Association

Every industry is riddled with examples of supposed experts missing the mark as they neglected to heed the voices of those they serve. Too often, those with the loudest megaphone attempt to explain what we should want as opposed to hearing what is truly desired or needed. Enter the Edsel, the Blimp, AOL, and the overwhelming majority of politicians.

So how do we avoid the pitfalls of leadership hubris? How do we ensure that customer voices are at the core of decision making? How do we balance the need for innovation with the ability to get novel products to gain traction in the marketplace? 

Humility and an intense interest in listening to and understanding the needs of business partners are crucial.  Agriculture at its very core is neighbors teaching neighbors how to make it greener, produce it healthier, grow it faster, or improve the taste. There are countless examples of successful firms engaging their business partners regarding new products, new approaches, and new directions. Your seed salesman, local coop, area veterinarian, regional sale barn are all forced to hear their customers and react accordingly. The feedback loop is quick, impactful, and brutal.

Other parts of our industry have been slower to seek input from commercial clients. One example is the area of Genetic Evaluation of Beef Cattle. The complexity of science, speed of innovation, overly-acronymed jargon are all reasons that well-intentioned folks sometimes isolate themselves and are slow to share findings and ideas with the end user. The historical breed association has been eager to report results, but less enthusiastic about incorporating commercial data or letting folks see behind the magic curtain of genetic analysis. Again, this isn’t necessarily an indication that associations were hiding or avoiding their obligation. Rather, in most cases it was a matter of limited personnel to share the message and limited scientific capability to incorporate commercial data.

Continue reading

 

Tuesday, 17 January 2017 11:08

A New Era of Objective Selection

Written by

RECENT DEVELOPMENTS IN GENETIC EVALUATIONS AND GENOMIC TESTING

Alison Van Eenennaam | University of California, Davis

The application of genomics to improve the accuracy of EPDs is a rapidly developing field. There are ongoing improvements in genotyping and sequencing technologies, statistical methods to increase the correlation between genomic predictions and true genetic merit, and the computing systems to handle the large datasets associated with animal breeding. One thing still remains true in the genomic age and that is the need to collect accurate phenotypic records. It is essential to ensure performance data, pedigree, and DNA information are recorded and reported accurately. Genomic predictions will only be as reliable as the data upon which they are based.  Although it might seem like the genomics era could signal the end of performance recording, the opposite is true. Now more than ever, it is important that producers accurately report data, and ensure that animals which are genotyped are correctly identified so that their information can contribute towards improving the accuracy of the genomic predictions of the future.

Continue reading

Tuesday, 17 January 2017 10:43

International Genetic Solutions Crossbreeding

Written by

AmericanCattlemen.com   |   Published March 14, 2016    |   

Cattle producers are always searching for answers to the question, “How do I improve my bottom line?” It’s a difficult question to answer, especially if, as a producer, you are making the decisions all on your own. The good news is there are organizations out there now that can off er cutting-edge tools to provide commercial producers with a wide range of data that focuses directly on the needs of the producer. International Genetic Solutions (IGS) is an unprecedented collaboration between progressive breed associations to enhance beef industry profitability. IGS is an outgrowth of listening to the commercial industry that began in 2010 when the American Simmental Association and the Red Angus Association of America came together to better support the profit goals of beef producers. Today, IGS is a collaboration of 12 progressive breed associations that have put self-interest aside to focus on the needs of the commercial cattle producer. These breed associations include: Red Angus, American Chianina, American Shorthorn, American Maine-Anjou, American Simmental, American Gelbvieh, Canadian Angus, Canadian Gelbvieh, Canadian Limousin, Canadian Simmental, Canadian Shorthorn and the North American Limousin Foundation. Will Townsend, Montana cattleman and Director of Commercial and Industry Operations with the American Simmental Association, is truly excited about what IGS is already doing for the cattle industry. “The goal is to promote good genetics and good breeding practices throughout the cattle industry,” says Townsend. “With over 16,000,000 total animals and 340,000+ new animals added annually, IGS has the largest multi-breed genetic evaluation system in the world and it gives commercial producers the tools to help them achieve the greatest profitability possible.” The goal of IGS is to provide genetics and genetic selection tools that maximize profit. This industry-leading partnership has made IGS the largest and most powerful genetic evaluation in the world. Its multi-breed capability is tailor-made for cattle producers who are serious about profit. The IGS and its collaborators have used science to develop the world’s most useful EPDs and profit-driven selection tools. This cooperation has allowed for unmatched research and industry-focused science. This large database and science-based research provides the best opportunity for commercial cattlemen to maximize profit through genetics.

Continue reading