Interbeef Guidelines

Version 2
InterBeef

contact: Interbull Centre - Dept. Animal Breeding and Genetic SLU,
Box 7023 S-75007 Uppsala, Sweden

Phone: 0046-18-671994
Fax: 0046-18-672848
E-mail : interbeef@slu.se
URL : https://www.icar.org/index.php/technical-bodies/working-groups/interbeef-working-group/

1. INTRODUCTION

Organisations willing to join InterBeef must be an ICAR’s client that has signed the Agreement for the Supply of Services with ICAR covering international genetic evaluation services for beef breeds and traits. This document is a description of the necessary actions and relevant files required by a new Organisations willing to join InterBeef Service.

2. THE ANIMAL INTERNATIONAL IDENTIFICATION

2.1 Format

A unique international identification for each animal is used in the international genetic evaluation system and will be referred to as the Animal International Identification (AIID). The AIID is constructed using the Interbull rules (referred as Interbull format in the document) and its structure is:

Breed + Country + Sex + Identification number

• The field ‘Breed‘ of the AIID refers to the breed of identification of the animal in its country of first registration (in most cases, the country of origin): 3 characters as defined in the Interbull Breed Codes, available at www.interbull.org (see Reference section).

• The field ‘Country’ of the AIID refers to the country of first registration of the animal (in most cases, the country of origin): 3 characters or 3 digits number as defined in ISO 3166 Codes, available at http://userpage.chemie.fu-berlin.de/diverse/doc/ISO_3166.html (see Reference section).

• The field ‘Sex’ of the AIID refers to the sex of the animal (M = male ; F = female)

• The field ‘Identification number’ of the AIID refers to the identification number of the animal in the country of first registration: 12 characters right justified with left blanks filled in with zero (“0”).

Ex: CHAFRAM006327826864 refers to

• A Charolais CHA
• First registered in France FRA
• A male M
• With the following identification in France 006327826864

Ex: BAQIRLF125693456875 refers to

• A Blonde d’Aquitaine BAQ
• First registered in Ireland IRL
• A female F
• With the following identification in Ireland 125693456875

2.2 Constructing the international identification number

An Organisation should first construct the international identifications for the animals born on its country. The general rule for this construction would be that a given identification number is unique within a country and a breed.

When an animal (live animal or issued from AI straw or embryo) is sold abroad, the Organisation in importing country should avoid re-registration; in case of re-registration the original ID and the national ID should be kept.

The animal ‘IE231112150014’ is, for example, a pure breed Irish limousine female named ‘SHRIFFTOWN RIPPER’. Ireland has constructed its AIID as followed: LIMIRLF231112150014.

The animal ‘FR0384106449’ is a pure breed French Charolais bull named ‘VLADIMIR’. That animal is also used in Ireland under the name ‘VLADIMER’ and the number ‘0384106449’ and has also been re-codified ‘108376849’ and ‘VLR’. France constructs its AIID as followed: CHAFRAM000384106449.

In the international identification number, hyphens, slashes, commas, dots, blank spaces or any other symbols are allowed.

The animal ‘UK9373 70-2/-7’, for example, is a UK Salers female animal and its international identification number would be ‘SALGBRF9373 70-2/-7’

3. IDEA - “Interbull Data Exchange Area” database at Interbull

Interbull has developed a database, IDEA (Interbull Data Exchange Area) that allows register users to all the pedigree and performances handling functionalities. IDEA is available at https://idea.interbull.org/
IDEA is exclusive for database users. To become a database user you need to belong to an Organisation participating to the Interbeef evaluation. If you want to join the Interbeef evaluation and therefore have access to IDEA, your Organisation needs to be associated to a username and a contact email address. To get your Organisation listed among the ones participating to the Interbeef evaluation, you have to: send an email to interbeef@slu.se providing information on:

• your Organisation: full name and address of the Organisation.
• the designated contact person: full name and email address of the person.
• Clearly state the breed(s) and trait(s) you wish to participate with. If your organization provides data as a joint evaluation including other countries, you clearly have to state for which other countries and breed(s) your Organisation provides pedigree and performance information.

Upon reception of your request, Interbull will send you by email a link to an IDEA-test environment together with a username and password so that you will be able to get acquainted with the IDEA functionalities and the user manual which is located under “Help”.

4. THE REQUIRED INFORMATION

An Organisation willing to join the Interbeef evaluation Service is expected to provide the following information:

• Pedigree information;
• Performance Information:
  1. Animals’ performance information for any traits the Organisation wants to provide data, within the ones currently evaluated (see Section 6.0 of Interbeef Code of Practice).
  2. Genetic parameters to be used in the international genetic evaluation.
  3. A list of animals’ IDs whose proofs are considered official for publication in the participating Organisation’s country.
  4. List of animals’ IDs for any ET (embryo transferred) animals. Such animals will receive an international breeding value based solely on their parent average. (This information is not mandatory).

4.1 Pedigree Information

The first step for an Organisation is to create a national pedigree file according to the format provided in section 4.1.1.
The created national pedigree file must be first checked with the CheckPedigree.py script provided by Interbull Centre (https://wiki.interbull.org/public/checkpedigree_python_instructions?action=print&rev=21).
The checking programs are available under the “Software” menu of the IDEA web interface.
The zip file produced by the Python script represent the only valid pedigree infile to be upload into IDEA database; no other files will be accepted by the system.
When the zip files created by the CheckPedigree.py script is uploaded to the IDEA database it undergoes a series of checks called “verification process” aiming at identifying the correct authoritative Organisation of each and every animals listed in the pedigree.
In general the system has been built in such a way that it accepts pedigree information only if these comes from the authoritative Organisation. Therefore, whenever you submit pedigree data, the system first checks that the combination “country-breed” in the animal ID matches with the combination your Organisation “owns”. If so, the records are considered correct and stored in the database. If this is not the case, then the records will be listed in the appropriate authoritative Organisation account waiting for verification, i.e. waiting for the authoritative Organisation to provide the correct pedigree information. In case the authoritative Organization is an Organisation that does not yet participate to the Interbeef evaluation, then the animals will appear as verified by Interbull until eventually that Organization would join the Service.
Example:
Ireland submits the following pedigree records to the database:

601 CHAIRLM458962315289 CHAIRLM789456123652 CHAIRLF369852147852 1985 OMAR IRL
601 CHAFRAM865412398745 CHAFRAM231658479851 CHAFRAF845693274125 1988 OSCAR IRL

In the first case the animal is an Irish Charolais sent by Ireland. Ireland is therefore the authoritative organization for that animal so the record gets stored in the database. In the second case the animal appears to be a French Charolais sent by Ireland. Ireland is no longer the authoritative organization for this bull therefore the system will send the record to the French account and will wait for France to provide its correct pedigree information. Then, the Organisation will have to correct its own database according to the feedback obtained from the IDEA database regarding verified pedigree information from authoritative Organizations. (see corrected foreign information in the IDEA user manual).

4.1.1 The Pedigree file format

The national pedigree file (601 file) must contain pedigree information of all animals included in the performance file in an animal-sire-dam format. Sires, dams and ancestors must also have an entry in this file as an animal; unknown ancestors should be coded ‘UUUUUUUUUUUUUUUUUUU’.

To ensure sufficient pedigree information it is recommended that pedigree information of animals born within a period equivalent to a minimum of three generation intervals should be included. The 601 file is a file issued per Organisation providing pedigree information according to the format shown in Table 1. The 601 files can contain one breed or multiple breeds.

Principles for the file preparation:

• Numeric information must be right justified.
• Characters must be left justified.
• All characters should be in upper case.

4.1.2 The Pedigree link file

The pedigree link file (202 file) has several functions related to the management of primary and alias animal identifiers (AIDs). In theory an animal should have only one identifier, the one corresponding to the country and herdbook of first registration (the international identifier or IID). However, due to limitations in national database systems and herdbook policies, animals have frequently been registered in different herdbooks with different identifiers. In order to uniquely identify animals across countries and herdbooks, it is necessary to manage multiple AIDs associated with an animal. In the Interbull pedigree database, an animal can have only one primary AID - which should be the IID - and any number of secondary AIDs (i.e. aliases, also called national IDs or cross references).
Invalid AIDs arise when importing countries enter the original AID incorrectly in their database, either due to typographical errors or omission of some characters in the alpha-numeric part of the identifier. These invalid AIDs can usually not be found in the originating country's database, which makes it impossible to verify the pedigree information of the animal in question. The pedigree link file serves three distinct functions:

• adding alias AIDs (eg. national IDs),
• linking two animals in the DB that are really the same animal, and
• marking an AID as invalid (i.e. not present in national herdbook database for the breed and country of first registration) by submitting AID2 as UUUUUUUUUUUUUUUUUUU.

The created link file must be first checked with the CheckLinks.py script provided by Interbull Centre (https://wiki.interbull.org/public/checklinks_python_instructions?action=print&rev=10).
The checking programs are available under the “Software” menu of the IDEA web interface.
The zip files produced by the Python scripts represent the only valid infiles to be upload into IDEA database; no other files will be accepted by the system.

4.2 Performance Information

Once the pedigree upload has been fulfilled then the Organisation will create the performance file according to the xml format described in section 4.2.1.
The created xml performance file must be first checked with the CheckPerformance.py script provided by Interbull Centre (https://wiki.interbull.org/public/CheckPerformance?action=print&rev=15) and available under the “Software” menu of the IDEA web interface. The tar file produced by the Python script represent the only valid performance infile to be upload into IDEA database; no other files will be accepted by the system. The tar file must be upload into the IDEA database as described here: https://wiki.interbull.org/public/IDEA_Beef_UserManual?action=print&rev=11.

Until the end of may 2020, it will be possible to create the xml performance file starting from the old flat format files (602, 603, 604 and 605). The format of old flat files is described in section 4.2.2. It will be possible to convert the old flat files into the xml file using the Convert Performance.py script as described here: https://wiki.interbull.org/public/performance_convert_program?action=print&rev=19.
Interbull Centre will be happy to provide assistance and support with the Convert Performance.py script, this in order to stimulate countries developing their own procedure on producing xml files.

4.2.1 Xml Performance file format

The XML file format used for performance data contains two distinct sections called traits and animals respectively.

The traits section corresponds roughly to the previous parameter file (i.e. 603 format) and the animals section contains the data from the previous performance file plus the data in the ET file and the officials files (i.e. 602 + 604 + 605 format).

The Traits Section

The traits section consists of a sequence of trait definition nodes. Each of those trait definition nodes has a header (the XML node) that defines the default values for some attributes of the trait (those attributes not usually specified per animal, i.e TYPE, EBREED, RCOU, mpe, etc) and a unique identifier to be able to refer to this trait definition from the animals (in attribute IDX). The trait definition node contains a sequence of nodes named "e" (short for environment effect) that defines ETYPE, COL and VALUETYPE for each environment effect that the trait consists of. The ETYPE attribute has to have a value of "R", "F" or "X" corresponding to random, fixed and covariate effects respectively. The COL attribute is the "name" of the effect and the VALUETYPE attribute specifies the type of the values for this effect.

The Animals Section

The animals section consists of a sequence of nodes named "a" (short for animal), each representing an animal, that have a header containing attributes id (for the AID), twin (for the TWIN value of this animal) and et (for the ET value of this animal). Each of the animal nodes contain a sequence of trait reference nodes (also called trait like the trait definition nodes). These trait reference nodes has a header containing attributes like herid, status and dep, but also the IDX attribute that explicitly references the trait definition with the same IDX attribute. Trait reference nodes contain a sequence of "e" nodes corresponding to the similar sequence of "e" nodes defined in the referenced (via the IDX attribute) trait definition. These "e" nodes simply contain the value of the environment effect with a ETYPE/COL/VALUETYPE of the "e" node in the corresponding position in the trait definition.

The special situation when an animal have no associated performance data, only ET status, is handled by having an animal "a" node without any contained trait reference nodes (i.e. only the header) and also without the twin attribute. There is a restriction that "a" nodes must contain at least one other node, this is handled by inserting a placeholder node "empty" under ET "a" nodes. For an explicit example, see the animal with id="LIMCZEM000575463142" below.

Example

The above description is perhaps made clearer by using a minimal example:

<interbull type="performance" subtype="interbeef" version="2.0">
  <traits>
    <trait TYPE="aww" EBREED="LIM" RCOU="CZE" ref="210" h2="11" cgn="1"
           twin="N" dam="Y" mpe="Y" pev="N" IDX="aww1">
      <e ETYPE="R" COL="HYS" VALUETYPE="Int"/>
      <e ETYPE="F" COL="ASEXTWIN" VALUETYPE="Int"/>
      <e ETYPE="F" COL="AACA" VALUETYPE="Int"/>
      <e ETYPE="F" COL="YEAR" VALUETYPE="Int"/>
    </trait>
  </traits>

  <animals>
    <a id="LIMCZEM000555163041" twin="1" et="0">
      <trait herid="CZE004200053400" dep="279.0" IDX="aww1">
        <e>0</e>
        <e>1</e>
        <e>3</e>
        <e>2012</e>
      </trait>
    </a>
    <a id="LIMCZEM000575463142" et="1">
      <empty/>
    </a>
    <a id="LIMCZEM000575163043">
      <trait TYPE="aww" status="05" Drel="35" Mrel="35" IDX="aww1"/>
    </a>
    <a id="LIMCZEM000123163044" twin="1" et="0">
      <trait herid="CZE004200053400" dep="222.0" status="05" Drel="55" Mrel="55" IDX="aww1">
        <e>0</e>
        <e>2</e>
        <e>3</e>
        <e>2002</e>
      </trait>
    </a>
  </animals>
</interbull>

In the above example, we can see a trait definition with IDX "aww1" that is of TYPE aww, EBREED LIM, etc that have four environment effects HYS/ASEXTWIN/AACA/YEAR.

In the animals section, the first animal (LIMCZEM000555163041) has a trait reference to that definition with herid, status and dep specified and the corresponding environment effects values 0/1/3/2012. The second animal (LIMCZEM000575463142) demonstrates the special situation when an animal have ET status, but no performance data. The third animal (LIMCZEM000575163043) shows when an animal have official status, but no performance data. The situation that an animal have both performance data and official status is demonstrated in the fourth animal (LIMCZEM000123163044).

Final Remarks

• The calving traits have an almost identical structure, the only real difference is the TYPE and IDX attribute values.

• The above description of the structure of the XML file for performance data is valid for autumn/winter of 2019, and replaces the previous version (used for 1904t).

4.2.2 Old Performance flat files format

• The 602 file representing the national performance file (limited until now to adjusted weaning weights, adjusted carcass weight and calving ease)
• The 603 file representing the parameter file.
• The 604 file representing the ET file
• The 605 file representing animals with official proof
• The Beef Form, see Reference section.

General principles for file preparation:

• Numeric information must be right justified.
• Characters must be left justified.
• All characters should be in upper case.

4.2.2.1 The performance file (602)

All animals with a performance must have a line as animal in the 601 pedigree file. The performance file must be constructed in agreement with the 601 file and be consistent with the international identification of the animal.
It is expected that countries participating with data support the ICAR guidelines for beef recording.
As indicated in the ICAR recommendations, any group of animals kept for the same purpose and at the same location shall be regarded as a whole herd. For a performance record to be considered an official record, the whole herd as defined above must be recorded. Each herd is identified with 15 digits identification, coded as characters:

• The first three digits correspond to the country, using the ISO 3166 Alpha-3 system, see Reference section.
• The next 12 digits correspond to the identification of the herd within the country. The original herd identification number should be used. The characters (letters or numbers) are right justified, the left blanks being filled with zeros.

All performance traits are associated with their environmental effects and complementary information (one observation line per measurement) as described in Table 2.

4.2.2.1.1 Special information for calving traits

Calving traits present a direct and a maternal component. For some traits, such as stillbirth and birth weight preparation of a file602 can be difficult because of the absence of an international ID for a dead calf. Given that Interbeef is an international data sharing platform, it is absolutely necessary that records are exchanged in a standardized way and that data quality must be ensured at the best of our capacity. Based on these two principles, there is a strong intention to keep phenotypic records under an universal formats capable of handling all possible traits (files 602 and 603). Likewise, only valid international animal identification is admitted into the system (file 601).

Phenotypic records for calving traits will therefore be reported as maternal traits (the International ID of ANIMAL in the file602 will refer to the dam ID) with the necessary information to fit a model for the direct trait added in the "environmental effects" fields of file 602. This includes additional information about the calf, such as sire of the calf, sex of the calf, calf birth date and calf ID (when available). Consider, for instance a record for calving ease (cae). In file 602, a typical record could be reported as shown in Table 2 and in session "Examples performance and parameter files" - calving traits

When calf additional information is reported in File602 (Table 2), the equivalent parameter File603 should also refer to the same variables. Although calf additional information is reported in the "environmental effects" field, it is not necessarily intended to be fit as an effect into the statistical model for the maternal trait. Since records on file 602 refer do the CALF DAM, in the parameter file the type of effects refer to the model for the maternal trait only. Therefore, variables that are reported in File602 to be used in the direct trait model should be labeled as "C" in the effect type field, so it is clear they are not fixed or random effects, neither covariates in the maternal trait model.

In the example in session "Examples performance and parameter file" for calving traits, calf sex, parity, season and age at calving are included as a fixed effects (effect type "F" in File603), while calf sire, calf birthdate and calf ID are labelled as effect type "C", which means they will not be fitted in the maternal trait model, but used for modelling the direct effect. For instance, calf ID, calf sire and calf birthdate would be needed to create a pedigree structure for the direct trait analyses.

4.2.2.1.2 File format performance file

Table 2: Performance file (602 file format)

^b Repeat this field n times (n = 1,…,NENV), adding one (1) empty space between fields; contemporary group should come as the first effect

4.2.2.2 The parameter file (603)

The parameter file (603) contains the name of two variables and the names of the environmental effects for each trait included in the performance file (602). This file must be provided together with the 602 file.

4.2.2.2.1 File format parameter file

• Table 3: Parameter file (603 file format) – one record per trait-breed combination

^b Repeat these fields n times (n = 1,…,NENV), adding one (1) empty space between fields; contemporary group should come as the first effect

Notes

4.2.2.2.2 Examples performance and parameter files

Animal weaning weight

Example performance file for aww:
602 AWW LIM CZE LIMCZEM000000181517 1 0 CZE008101626410        266   4                13118                    1                    3                 1995

The related paramter file (603) to the performance file (602) above should look like this:
603 AWW LIM CZE 210   11  1          N Y Y N 4   CG                  R ASEXTWIN             F Age_of_Dam           F Birth_Year           F

Explanation 602 file for aww:
602 = record type; AWW= trait; LIM=breed; CZE= country sending information; LIMCZEM000000181517=aid; 1=twin_info; 0=et_info ;CZE008101626410= herdid; 266=dependent_variable; 4= number of env. effects (should be the same as the number of env. effects in the parameter file (603);  13118=env. effect 1 which should be related to the env. effect 1 in the 603 file (in this case HYS), 1= env. effect 2 which should be related to the env. effect 2 in the 603 file (in this case ASEXTWIN); 3= env. effect 3 which should be related to env. effect 3 in the 603 file (in this case Age_of_Dam); 1995= env. effect 4 which should be related to the to env. effect 4 in the 603 file (in this case Birth_Year)

Explanation 603 file:
603 = record type; AWW=trait; LIM=breed; CZE= country sending information; 210=reference age or class; 11= trait heritability; 1= Minimum number of observations per CG; N=twin info; Y=Maternal (genetic) effect fitted in the model ; Y=Maternal permanent env. effect fitted in the model; N= Permanent env.effects included in the national model; 4= number of env. effects (should be same number as in the performance file 602); HYS=env.effect1 which is related to the env.effect 1 in the performance file; R= How env. effect is fitted into the model(type of effect); ASEXTWIN=env. effect2 which is related to the env.effect 2 in the performance file; F=how env. effect nr 2 is fitted in to the model; Age_of_Dam=env.effect3 which is related to env effect 3 in the performance file; F=how env. effect nr 3 is fitted in to the model; Birth_Year= env effect nr 4 which is related to the env effect 4 in the performance file; F=how env. effect nr 4 is fitted in to the model

Cavling traits

Example performance file for cae:
602 CAE LIM CZE LIMCZEF000000260705 1 0 CZE008101626410       1      8                18256 LIMFRAM001984004136  1                                19950401 LIMCZEM000000181517                  1                    3                    1995

Explanation 602 file for cae:
602 = record type; CAE= trait; LIM=breed; CZE= country sending information; LIMCZEM000000181517=aid (referring to the dam of the calf); 1=twin_info(related the the calf); 0=et_info (related the the calf) ;CZE008101626410= herdid; 1=dependent_variable; 8= number of env. effects (should be the same as the number of env. effects in the parameter file (603);  18256=env. effect 1 which should be related to the env. effect 1 in the 603 file (in this case HYS), LIMFRAM001984004136= env. effect 2 which should be related to the env. effect 2 in the 603 file (in this case CASI); 1= env. effect 3 which should be related to env. effect 3 in the 603 file (in this case CSEX ==> 1=Male, 2=Female,3=Unknown); 19950401= env. effect 4 which should be related to the to env. effect 4 in the 603 file (in this case CABI); LIMCZEM000000181517 =env.effect 5 which should be related to env.effect 5 in the 603 file(in this case CAID ) ; 1= env. effect 6 which should be related to the env.effect 6 in the 603 file (in this case ASEXTWIN); 3=env.effect 7 which should be related to env.effect 7 in the 603 file(in this case Age_of_Dam); 1995 = env.effect 8 which should be related to env. effect 8 in the 603 file (in this case Birth_Year).

603 CAE LIM CZE 1     9   1          N Y Y N 8   CG                   R CASI                 C CSEX                 C CABI                 C CAID                 C ASEXTWIN             F Age_of_Dam           F Birth_Year           F
Explanation 603 file for cae:
603 = record type; CAE=trait; LIM=breed; CZE= country sending information; 1=reference age or class; 9= trait heritability; 1= Minimum number of observations per CG; N=twin info; Y=Maternal (genetic) effect fitted in the model ; Y=Maternal permanent env. effect fitted in the model; N= Permanent env.effects included in the national model; 8= number of env. effects (should be same number as in the performance file 602); HYS=env.effect1 which is related to the env.effect 1 in the performance file; R= How env. effect is fitted into the model(type of effect); CASI=env. effect2 which is realted to the env.effect 2 in the performance file; C=how env. effect nr 2 is fitted in to the model; CSEX=env.effect3 which is related to env effect 3 in the performance file; C=how env. effect nr 3 is fitted in to the model; CABI= env effect nr 4 which is related to the env effect 4 in the performance file; C=how env. effect nr 4 is fitted in to the model; ASEXTWIN=env.effect nr 6 which is related to the env.effect 6 in the performance file, F=how env. effect nr 4 is fitted in to the model;  Age_of_Dam=env.effect 7 which should be related to env.effect in the performance file; F=how env. effect nr 4 is fitted in to the model; Birth_year= env. effect 8 which is related to env.effect 8 in the performance file; F=F=how env. effect nr 4 is fitted in to the model

4.2.2.3 ET file (604)

The ET file (604) contains a list of ET IDs which will be included in the pedigree used for the international evaluation. Country willing to get international breeding values for ET animals based on parent average will have to provide the file at each evaluation. This file is not a mandatory file. Please note: For calving traits, the embryo transfer field in the 602 file is related the the calf and not the the dam itself.

4.2.2.3.1. File format ET file

4.2.2.4 OFCL file (605)

The OFCL file (605) contains a list of IDs of animals with official proof in the participating country. According to a new proposal for the publication policy, if the national proof of an animal is considered as official, the international predicted breeding value of the animal on the scale of that country would be publishable.

4.2.2.4.1 File format maternal and direct Official file

4.3 Crossbred Animals

At the combined Interbeef Technical- and Working group meeting in Padova (Italy, October 2018), it was decided to include crossbred animals in the animal weaning weight evaluation for all breeds of evaluation (HER, AAN, SIM, LIM and CHA). This means that performance records for crossbred animals in the performance file will be included in the international genetic evaluation. A calf with sire breed = breed of evaluation (i.g LIM, CHA, SIM, HER or AAN) and dam breed = any other breed is considered to be cross breed and should have breed code XXX (i,g XXXIRLM123456122333). The pedigree records for the crossbred animals should be uploaded in IDEA as usual.

5. ABBREVIATIONS

Table 6: List of abbreviations used in Interbull files and documentation

REFERENCES

ISO 3166 Country Codes: http://userpage.chemie.fu-berlin.de/diverse/doc/ISO_3166.html
Breed Codes for International Genetic Evaluation of dairy and beef cattle: https://wiki.interbull.org/public/breed%20codes?action=print&rev=19
Guideline to Form Beef: http://www.icar.org/wp-content/uploads/2015/12/Interbeef-Form-BEEF-guideline.pdf
Form_Beef: http://www.icar.org/wp-content/uploads/2015/12/Interbeef-Form-BEEF.pdf