In pvrqualitasag/PedigreeFromTvdData: Import pedigree data from TVD-Input files

Disclaimer

This vignette describes concepts, strategies and systematics developed for running data consistency checks for pedigree data.

Introduction

From a theoretical point of view, pedigrees are Directed Acyclic Graphs (DAG), where a graph $G$ is defined by the set of vertices or nodes $V$ and the set of edges $E$. The set of vertices correspond to the set of individuals in the pedigree. Edges correspond to directed relationships from parents to their offspring. In a pedigree, the direction of the edges is always from parents to offspring. A pedigree must not contain any directed cycles, i.e. there cannot be any path corresponding to a series of directed edges linked together where a parent appears after one of its offspring. These two properties

1. directed edges between parents and offspring
2. no cycles

together have led to the name DAG.

In a directed graph, we can distinguish between the number of edges that are coming into a node and the number of edges going out of a node. The former number is called in-degree and the latter number corresponds to the out-degree. For a pedigree, the maximum in-degree for every node is $2$.

Based on the described properties, we can define a set of consistency requirements that must be fullfilled when a pedigree is constructed.

Consistency Requirements

This section describes consistency requirements that are derived from the properties of a pedigree described in the previous section.

Properties of a DAG

The following list of requirements is derived from the properties of a DAG

• Uniqueness of individuals: Considering the fact that a pedigree is a DAG and the definition of the nodes as a set, it follows that nodes must be unique. Hence every individual can only appear once in our pedigree. As a consequence of that, we can use the IDs associated to the nodes as unique mean of identification. In the database-world this corresponds to a primary key, meaning that a primary key uniquely identifies a given record.

• No cycles: The property of a pedigree being acyclic was already described in the previous section

• In-degree of nodes: The maximum in-degree of every node is $2$.

• Parents older than offspring: Parents must be older than offspring

• Parents must have the correct sex

Other requirements

There are additional properties which are more related to data-processing issues. Those issues mostly involve the correctnes of certain data-formats.

• correct formats of IDs for individuals and parents
• correct formats of birthdates

Implementation

Implementations do depend among many things on the type of data representation of a given pedigree.

Data representation

One of the most commonly used data representation of a pedigree is the so-called node-list or adjacency-list. This is a tabular list with columns containing IDs for animals, IDs for parents and additional information such as sex and birthdate. One row of the list corresponds to the available information for a given individual and hence must be unique. Such a row is also called a pedigree record.

Types of implementation routines

When it comes to verifying the consistency requirements two type of implementation routines can be imagined

• checks are routines which take a pedigree and a condition given by a specific requirement. The only thing they do is they search the pedigree for entries which do not fullfill the consistency condition. As a result, the check-routines return the set of primary keys which do not fullfill the conditions used in the check. Besides the set of primary keys, other useful information might also be returned by the check-routines. By convention, the returned result of the check-routines are always in the format of a tbl_df where the first column contains the primary keys. From an implementation design point of view it is important to note, that the check-routines work without any side-effects on the pedigree, i.e., they do not change the pedigree.
• transformations take a pedigree as input and give as output a different pedigree where one of the following two operations were applied
  • invalidate or delete a complete pedigree record or
  • invalidate or delete a field of the pedigree record

The descriptions of each check is described in a companion vignette on Pedigree Checks - Implementations which is also available in this package.

The descriptions of each transformation is described in a companion vignette on Pedigree Transformation - Implementations which is also available in this package.

Vignettes Overview

• Reading Fixed Width Formatted Data: comparison of different methods for reading large amounts of data.

• Pedigree Checks - Implementations: description of each implemented check-function.

• Pedigree Consistency Checks: time and memory requirement for some check-functions.

• Strategies for Pedigree Transformations: description of different method to delete record or invalidate a field.

• Pedigree Transformation - Implementations: description of each implemented transformation-functions.

• Main Pedigree Build: description of implementation of each check-funtion and transformation-function in one function to call TVD-Data.

• CheckList for TODOs: checklist to keep in mind by merging programmation.

• User Manual: Manual.

Session Info

sessionInfo()

Latest Update

r paste(Sys.time(),paste0("(", Sys.info()[["user"]],")" ))

pvrqualitasag/PedigreeFromTvdData documentation built on May 29, 2019, 7:50 a.m.