Reflections on Pedigree Analysis

by C.A. Sharp

First published in Double Helix Network News, Winter 2008   Rev. Nov. 2013


I have been performing pedigree analysis for health traits in Australian Shepherds since the early 1990s.  The scope of the effort has broadened and I have refined the technique throughout that period to better meet breeders’ needs.  In 2008 it became a program of the Australian Shepherd Health and Genetics Institute (ASHGI) as part of the organization’s planned International Directory for Australian Shepherd Health but much of the work still had to be done essentially by hand which limited the number of clients that could be served.  In 2013 the pedigree analysis service became web-accessible and most of its operations computerized, allowing clients fast, accurate service without the need to wait for an appointment.

The result of all this growth is a system that may seem complex and daunting to the uninitiated.  It is my hope that this article will make the process more comprehensible to breeders trying to interpret her dog’s results as well as inform those who want to know more about the process before  deciding whether they want to use it.

 Managing the unavoidable

Unwanted genes are a fact of life for breeders; even the very best dogs have them.  The point is to know as much as you can about your dog’s genetic potential for the unwanted traits so you can make informed breeding decisions.  The service has never been intended to be a culling mechanism for a breeding program; no dog should be withheld from breeding on the basis of pedigree analysis scores alone.  The IDASH Pedigree Analysis Service (PAS) offers a tool you can employ to make the best possible decisions regarding unwanted health issues, as well as a few other traits like dental conformation and unrecognized coat colors.

Prior to the advent of DNA tests, we often had no idea what a dog’s genotype for a particular trait might be.  Since the tests available are still limited, this continues to be the case for most traits.  PAS searches the pedigree of a given dog for connections to traits of interest to determine the dog’s relative risk of having genes any of approximately two dozen traits which it might, in turn, pass to its offspring.   The higher the score for any given trait, the greater the risk.

Some might feel that this would be a good way to decide what dogs not to breed.  This is not a good way to use the reports because a score only represent the amount of potential for a trait, not the certainty that the dog actually has those genes.  If you eliminate from breeding all dogs of undetermined genotype that have a high scores for something you don’t want you risk unnecessarily discarding all the desirable genes those dogs might have.   The goal should be to breed dogs with problematic scores to mates with better scores, thereby reducing the level of risk for their pups.

This is essentially what breeders already do when selecting away from faults in conformation or important behavioral traits.  The PAS gives breeders an objective measure for various health and disqualifying color traits that they can consider in their overall decision-making process.


It is helpful to know how the system developed.  Shortly after the successful conclusion of the initial study of Collie Eye Anomaly (CEA) in Australian Shepherds in 1991, it occurred to me that the data I had gathered for that study could be used to help breeders reduce the risk of what was then considered our major genetic health issue.

For a number of years CEA was the only trait I addressed.  The process was cumbersome, with the data recorded on 3×5 cards that I had to thumb through every time I did a pedigree.  Reports were typed one at a time.

Years passed; I got a PC and a good pedigree program.  A home-office reorganization project reminded me that I had accumulated volumes of pedigree data on other genetic issues which could be used in a similar manner once I entered the data into my pedigree database.  In 1999 I began reviewing pedigrees for about 18 different traits.  Over the years I have added a few more traits and adjusted the scoring system to make it the best possible reflection I could devise of a dog’s family history for a given trait without revealing specific confidential information.

Beginning in January 2008, the program ceased to be my private project and became part of ASHGI’s new IDASH program.  At that point, I continued the processing and preparation of reports much as I had before, but moving the program into IDASH will guarantee that no matter what might happen to me, the data would not be lost and the program could continue.

Until 2013 the data remained in my pedigree program with the review and calculations are done largely by hand.  This was time-consuming (each pedigree took about an hour) and prone to clerical error.   ASHGI engaged a computer programmer to automate the service, leading to a computer-based system.  Some paperwork, mostly verifying ownership and reviewing health documents, has to be done by a human, but the rest is computerized rendering clerical and computation errors a thing of the past.

 Data Sources and Quality Control

The information comes from multiple sources.  Most of it was presented to me, or more recently to ASHGI, by concerned owners and breeders who are grappling with health issues in their dogs.  This is supplemented with the public domain information from open health registries and listings in the United States and Europe.

PAS has been criticized because the data is not complete:  There is no mandatory reporting and therefore not all dogs that have a given trait are listed.  Some fault it because owner-submitted data may sometimes be inaccurate.  These are legitimate arguments and deserve response.

It is true ASHGI does not receive information from everyone who might have a dog affected with one of the traits we track.  Frankly, the same can be said for OFA, CERF, or any voluntary health registry.  Even the mandatory registries in Europe typically receive data on potential breeding dogs, not every registered dog produced,   Even though our data may have gaps, it is better to make use of the information available than proceed in total ignorance.

Owner-submitted data may sometimes be in error, particularly for diseases that are not simple to diagnose.  Submissions are screened and additional questions asked where clarification is necessary.

It is extremely unlikely that someone would deliberately pump false information into the system. The owner-derived data comes from people who contact ASHGI wanting to know more about traits that have been diagnosed in their dogs or which they are trying to manage in their breeding programs.  These people have no motivation to give information that would make the situation appear worse than it actually is.   Should information appear questionable and we cannot resolve those questions, it doesn’t become part of the database.  Should we find that something in the base is in in error, as when a diagnosis is later changed, we will correct the error.

Public-domain data derived from health registries is the “gold standard” for information and is used whenever it is available.  The most useful data source, a veritable “platinum standard,” is DNA screening test results.  These tests reveal the actual genotype of the dog.  It doesn’t matter what potential risk lurks in that dog’s pedigree when you know, for certain, what version(s) of a particular gene it has.

 Traits tracked

PAS evaluates a range of traits, including major and minor health issues, dental faults and disqualifying coat colors.  Initially I listed only the more common health and dental issues in the breed. As time went by, I added disqualifying color traits and less-common health traits that could have serious or fatal consequences.  Recently we have included Items for which DNA tests are available for Aussies.

PAS also determines the coefficient of inbreeding (COI) for each pedigree analyzed.  The COI is not a direct indicator of health status, but dogs with higher COIs are more likely to have inherited two copies of disease-causing gene variants from ancestors which appear on both sides of the pedigree. High COIs may also indicate a risk of inbreeding depression if they have been maintained over several generations.  [For more information on how to interpret and utilize COIs, click HERE.]

 The Process

I had to develop a reasonable and consistent scoring system that could be applied to a group of traits with varying and undetermined modes of inheritance.  I settled on a modified percentage of ancestry calculation.  Breeders use percentage of ancestry to determine how much a particular ancestor contributes to a pedigree.  The result is expressed as a percentage.  For example, if a dog’s only appearance was as grandsire of the subject, its percentage would be 25, but if it was the grandsire on both sides the percentage would be 50.

Instead of seeking repeated names in a pedigree, I look for connections to a particular trait.  The resulting scores are point values, not percentages.  Percentages would require that we know the exact mode of inheritance, which we usually don’t.  It would also require having a different type scoring system for each trait.  That would be both cumbersome and confusing, so I settled on a system that could be applied to any given trait and was simple enough that any breeder who was so-inclined and had a good data set could do it herself.

Initially, scores range from zero to 100, but the use of a 100 point scale inevitably lead people to think of the scores as percentages.  Therefore in 2010 a 10 point scale was adopted.  PAS calculates and records the scores to one decimal point but on the reports they are rounded to the nearest whole number.

Scores indicate whether a dog has more or less background for that trait.  A zero score will indicated on the report as NKB (no known background) or, for those items with DNA tests, in a manner that reflects its test status.  NKB is used in recognition of the fact that our data is not complete and it is possible that the dog may have background of which we are not aware.

Since the frequency of these traits varies considerably, the score alone might not be sufficient for breeders to make sound decisions.  Starting in 2007 I began logging all the test results so I could develop averages for each trait.  The upgraded and web-mounted PAS will offer scores that reflect variance from these averages.  Individual breeders are unlikely to have the amount of data necessary to do this, but using the straight scoring method will enable them to make comparisons among their own dogs and specific dogs they may be interested in. 

Reflections on PA -CalculationsPAS reviews the pedigree seeking ancestors who had the trait or were the parents or grandparents of a dog that had it.  This incorporates breadth of pedigree while working through direct ancestors.  Parents are given half the weight of an affected and grandparents a quarter. PAS also notes in which generation these dogs appear. Each generation farther behind the subject dog is given half the weight of the one before.  If there has been no indication of a trait in five generations, it is extremely unlikely that the dog being analyzed will have inherited genes for it.   [See score chart.]

A score of 10 is equivalent to having an affected parent, 5 to having a parent that has produced the trait, 25 a grandparent that has produced it, etc. In most cases, however, more than one dog contributes to the result.  One could, using this method, have a dog that scored 20 (affected itself, two affected parents, etc.) but few dogs score more than 10.  People are comfortable with 10 point scales and 10 already indicates a very significant level of risk, therefore we set the ceiling at 10.

While the method of calculation is established, the system is designed to adjust to new information.  In 2007, evaluation of the data from ASHGI’s Cancer Survey revealed that hemangiosarcoma and lymphoma are very common and often follow family lines.  Therefore, those cancers were added to the list of traits in 2008.  Another change, the discovery that the mutation causing most cataracts in Aussies is a dominant, resulted in adjustment of relationship categories for that disease where the genetic status of a dog has become known.  The system will be able to adapt to future discoveries as well.


Pedigree analysis results are intended to inform the breeder about potential risks in her dog’s pedigree.  Even a very high score does not mean a dog will produce a particular disease or fault.  However, the higher the score the more likely this could happen so caution is advised.  It is also very important to consider variance from average.  If the average score is high, as it is for epilepsy, most dogs will have a high score.  What the breeder must do is consider how much higher than average it is when making decisions about whether to breed or what mate to breed to.  In a similar matter, if the average score is very low, being a few points above average may not confer significant risk though, ideally, the breeder would want to choose a mate that would lower the score, especially if the health issue is potentially fatal or has serious quality of life issues.

Ideally, you would like to see all individual trait scores under 3 prior to calculating variance from average , but this is very rarely the case.  A score of 3 or higher is reason for concern.   You can determine this score when viewing a PAS report by adding or subtracting the trait variance (depending on whether it is above or below average) from the average score given in the report.   The higher the base score, the more concerned you should be. If you plan to breed the dog, try to select mates that will lower the score in their mutual pups. A litter’s score will not exceed an average of the scores of the two parents.  (It may even be lower because data in the parents’ 5th generation “drops off” the litter’s pedigree.) Avoid crosses that score in excess of 50 for any serious disease if at all possible but recognize that where average scores are high this may not be possible except over the course of several generations.  If your dog exceeds the average score on one or more traits, attempt to reduce those scores in its puppies below the average whenever possible.

If a dog scores 3 points or more for a trait where DNA testing is available, the dog should be tested so its genotype is known and appropriate mating decisions can be made.

For serious health traits that typically don’t appear until the dog is a few years old, if the score is 3 or greater you may want to hold off breeding that dog until it is past the typical age of onset for the trait.  Delaying breeding will reduce the chance that the dog may become affected after it has produced offspring.

If multiple traits are a concern in any particular dog, you are unlikely to be able to reduce risk on all of them at once. Set priorities based on health impact:  Epilepsy is more serious than cataracts and both are more serious than missing teeth or too much white.  But if your scores for epilepsy and cataracts are low while your scores for missing teeth and excess white are high, then the higher score items should be of greater concern.

Pay attention to the current average scores.  A score of 5 isn’t good, but if the average the trait is high (the 2011 epilepsy average was 5) 5 might be better than most or a significant portion of the breed.   A dog may have one or two very high scores and still have a total score that is below average.  The overall health risk for such a dog is better than most.  You will have to be careful about those high-risk items, but the picture may not be as gloomy as it appears on first inspection.

Another point on averages:  If a dog’s score is 7 and the average is 1, that indicates the dog carries tremendously more risk for that trait than the majority of the breed.  The good news here is that there will be plenty of potential mates with whom the dog is unlikely to produce the trait.

When looking for a stud to use with your bitch, or reviewing records on a bitch presented to your stud, you will want to seek crosses that reduce risk in the offspring wherever possible.  Do not issue a blanket refusal to breed to any animal that has a higher score on something than yours does.  Mate selection is a balancing act:  Look at all the plusses and minuses on each side, not only with PA scores, but with structural and behavioral traits of importance.  Do the potential mates complement each other when all angles are considered?

Because new information flows into the database continuously, it is advisable that pedigree analysis be repeated about every three years for any breeding dog that is still living or for which semen is being stored for future use.  One can consolidate analyses to some degree for full siblings.  Their scores will be exactly the same unless one has produced something another has not or has DNA test results that differ from its siblings’.


 A note on how to analyze the good stuff

The PAS system works for things you do not want, but of course every breeder wants to know how to maximize potential for good traits.  For this I recommend Dr. Carmen Battaglia’s “stick dog” pedigree analysis.    Details can be found on his website,

Essentially, through the use of a 3-generation pedigree with color-coded stick figures you can readily identify the strong and weak points in your dog’s pedigree.  The color coding indicates whether some aspect of the dog was excellent, good, fair or poor.  Notations are added  for detail (“excellent topline,”  “somewhat straight shoulder,” etc.)  You can tell at a glance if there is strong background for the traits you desire.

The system is designed for tracking conformation but could be adapted for behavioral traits desired in work or performance events.

Without knowledge of your dog’s genetic potential – both for good and for ill – you cannot make informed breeding decisions.  If you are flying blind when it comes to health issues, you will produce them more often than if you know something could go wrong and you take steps to avoid it.  ASHGI’s IDASH Pedigree Analysis Service is one more tool you can use toward that end.