Eildert Groeneveld

Simultaneous estimation of the covariance structure of traits from field test, station test and litter recording in pigs

Abstract The covariance structure was estimated by REML for a joint genetic evaluation of traits from field test, station test and litter recording with data from the Czech Landrace (LA, 86 833 records) and the Slovak White Meaty breed (WM, 58 348 records). Production traits from the field test (average daily gain [ADGF], lean meat percentage [LMF], and backfat thickness [BFF]) and from the station test (average daily gain [ADGS], valuable cuts [VCS], and backfat thickness [BFS]) as well as reproduction traits (number of piglets born alive in the first litter [NBA1] and number born alive in the second and higher litters [NBA2+] as a trait with repeated measurements) were analyzed jointly in a six-trait (LA) or seven-trait (WM) animal model. The estimated genetic correlations between NBA1 and NBA2+ (0.83 for LA and 0.78 WM) support the decision to treat first and further litters separately in genetic evaluation. The proportions of variances and correlations were also estimated for nongenetic effects in the models for both breeds (herd or herd–year–season, station–year–season, litter effects, and permanent effect of sow). For ADGF, a very high proportion of variance (0.43) was observed in WM for the herd–year–season effect and in LA (0.27) for the herd effect. The herd–year–season effects for reproduction traits were low (0.05–0.06 in WM, 0.01–0.02 in LA). The estimated genetic correlations between production and reproduction traits were small with the highest absolute values between NBA2+ and ADGS (−0.12 for LA and −0.28 for WM). As not all of them were zero, a separate genetic evaluation of reproduction would be incorrect. Also, because of the low heritabilities of the reproduction traits any additional information will be useful making a joint evaluation of production and reproduction traits preferable. Covariance estimates and parameters for genetic and nongenetic effects in the models were widely discussed with special attention to the most recent literature sources.

An adaptable platform independent information system in animal production: framework and generic database structure

Abstract Currently, no published strategy exists for the creation of integrated databases for individual animal recording systems as required for herdbooks and breeding programs. As a result integrated databases are either created at very high costs or—more commonly—not at all, which is a severe restriction for the implementation of modern animal breeding tools. A development and migration strategy for databases storing individual animal records is presented. A generic data model has been developed which can be used across species and animal identification and recording systems. For implementation of a new information system the procedures for its adaptation are outlined. A software framework is made available under the open source model using the GNU public license (GPL). Thus, information systems can be developed and implemented without any software license costs. The procedures presented will allow the creation of integrated information systems required where this has been so far not possible for cost reasons. Furthermore, the adaptable platform independent information system (APIIS) platform may serve as a basis for further cumulative development of application software in animal breeding and production management. The software is available at http://www.apiis.org .

CryoWEB: Web software for the documentation of the cryo-preserved material in animal gene banks.

Many countries in the world have set up national domestic conservation programmes and collections of long-term storage of cryo-preserved animal genetic material. We have developed a web based software ‐ CryoWEB for the documentation of such collections. The software is generally applicable to all livestock breeds, independent of the donors species and the type of preserved material. The software can record basic herd‐book information for the donor animal, the structure of the storage facilities, description of the stored samples and their distribution within the gene bank. It also traces the movements of the sample vessels within the storage facilities and the usage of sample units. The outputs implemented in CryoWEB address the informational needs of the gene bank manager in her everyday tasks. Availability CryoWEB is publicly available athttp://cryoweb.tzv.fal.de

Integrated data aquisition, transmission and processing systems in two pig breeding schemes in the Federal Republic of Germany

Abstract Selection in breeding programs requires complete information on all activities in the population. Since selection decisions are largely based on the results of statistical calculations carried out by computers, information has to be available in computer compatible form. Two operating systems are described. The first deals with the aspects of data flow and analysis in the Federal Hybrid Crossbreeding Scheme for pigs and has been in commercial operation since 1980. The second is based on routine manipulations of data from the pig breeding herdbook societies, comprising all of the station tests and 40 per cent of field tests of boars in Germany. Data collection is carried out under program control using microcomputers. Data are transmitted via the public telephone network, employing a LSV 2 protocol to ensure safe data transmission. Data base management systems are used for data storage and retrieval. Much emphasis has been put on system flexibility to speed up program development and to ease future changes in data and program structure.

Performance characteristics of different solving strategies in multivariate mixed models

Abstract Scope and optimization strategies for solving mixed linear models with PEST (general purpose package for multivariate prediction and estimation) are investigated on the basis of three multivariate problems with system sizes of 4116, 30 016 and 472 252 equations. A univariate model with 1 813 168 equations was run to assess the performance of PEST with large problems. Where a sub-matrix of the mixed model equations was solved simultaneously computing time was reduced to around 1% if compared with complete Gauss-Seidel iteration on data. With the possibility of different incidence matrices for each trait in multivariate models no precorrection of data is required but regressions can be added to the model without much penalty in processing time. Overrelaxation may improve the rate of convergence by 25% and setting a “ceiling” for the number of iterations for the simultaneous part reduced the CPU time by 11%. Cholesky transformation further reduced CPU time between 30 and 50% depending on the required accuracy. The largest system was an animal model for milk yield of nearly 900 000 cows resulting in 1.8 million equations which required 26 Mb of memory during the solving process taking 37 min per round on a SUN 4 workstation.

An open information system for the swine production and marketing industry: Its scope, topology and telecommunications strategy

On the basis of data originating from producers and users in the swine production industry, the structure of a comprehensive information system (IS) is proposed that should cover the information needs of all participants in the process. Some present-day computer/record keeping systems are outlined and their limitations pointed out. From considerations of the scope of a comprehensive IS, it is concluded that the resultant topology of an IS calls for on-farm computers handling all data originating and pertaining to their respective locations and a central system which stores all data from every participant. The central database of identical structure contains all peripheral data as copy of the onfarm database. Also, the same operating system is run throughout the network. The information system is run functionally as a distributed database. The ensuing problems connected with redundant central storage of data are evaluated leading to the development of a high-level telecommunication protocol. A telecommunication strategy developed and implemented in the context of a distributed herdbook management system is presented from the perspective of a system designer and discussed in relation to the problems in an open information system in pig production. Its limitations are pointed out leading to a telecommunication strategy based on UNIX and international standards.

Strategic considerations in the choice of basic software and hardware for information systems in animal production.

Abstract Prior to implementing any information system (IS), decisions have to be made about the basic software and hardware to be used. The choices greatly influence efficiency of implementation and future flexibility. Strategic goals developed are: hardware independence, future viability, expendability, operational safety, efficiency of programming and maintenance, and end-user evaluation. Choices are discussed on the basis of the established goals. As a result, a relational database was chosen. Application development was done using high-level tools, based on the non-procedural language SQL. With the rapid development of hardware over the last year, it is now possible to run to operating system UNIX on hardware in the PC class. This allows the use of a uniform and open operating system and identical application software on all computer systems irrespective of their size, ranging from personal computers to supercomputers.

A database for efficient storage and management of multi panel SNP data

The fast development of high throughput genotyping has opened up new possibilities in genetics while at the same time producing immense data handling issues. A system design and proof of concept implementation are presented which provides efficient data storage and manipulation of single nucleotide polymorphism (SNP) genotypes in a relational database. A new strategy using SNP and individual selection vectors allows us to view SNP data as matrices or sets. These genotype sets provide an easy way to handle original and derived data, the latter at basically no storage costs. Due to its vector based database storage, data imports and exports are much faster than those of other SNP databases. In the proof of concept implementation, the compressed storage scheme reduces disk space requirements by a factor of around 300. Furthermore, this design scales linearly with number of individuals and SNPs involved. The procedure supports panels of different sizes. This allows a straight forward management of different panel sizes in the same population as it occurs in animal breeding programs when higher density panels replace previous lower density versions.

An efficient approach to the deployment of complex open source information systems

Complex open source information systems are usually implemented as component-based software to inherit the available functionality of existing software packages developed by third parties. Consequently, the deployment of these systems not only requires the installation of operating system, application framework and the configuration of services but also needs to resolve the dependencies among components. The problem becomes more challenging when the application must be installed and used on different platforms such as Linux and Windows. To address this, an efficient approach using the virtualization technology is suggested and discussed in this paper. The approach has been applied in our project to deploy a web-based integrated information system in molecular genetics labs. It is a low-cost solution to benefit both software developers and end-users.