American Journal of Medical Genetics 36:155-160 (1990)
Occasional Paper A Visual Interface to Computer Programs for Linkage Analysis Cyril J. Chapman Department of Laboratory Seruices, Wellington Hospital, Wellington, New Zealand
This paper describes a visual approach to the input of information about human families into computer data bases, making use of the GEM graphic interface on the Atari ST. Similar approaches could be used on the Apple Macintosh or on the IBM PC AT (to which it has been transferred). For occasional users of pedigree analysis programs, this approach has considerable advantages in ease of use and accessibility. An example of such use might be the analysis of risk in families with Huntington disease using linked RFLPs. However, graphic interfaces do make much greater demands on the programmers of these systems. KEY WORDS: user interface, pedigree analysis, linkage analysis
INTRODUCTION Any clinical geneticist who has attempted to analyze a human pedigree for genetic risks using either LIPED [ o t t , 19741 or the LINKAGE package [Lathrop and Lalouel, 19841, will be aware of the considerable difficulties experienced by anyone who uses these programs only occasionally. The methods used for data entry in these programs are not intuitive for a clinical geneticist, and data verification after entry into the computer files is difficult. Anyone who has access to a microcomputer system that uses a graphics-based interface with windows, menus, icons, and a pointer (mouse) will know that such difficulties are not a necessary part of computing. This paper describes a program that uses the graphic interface provided by GEM on the Atari 520ST and the IBM PC AT. A similar data-entry system could be developed on a n Apple Macintosh or any of the other graphical interfaces now available, such as the OSi2 Presentation Manager or Microsoft Windows. Received for publication November 4, 1988; revision received October 11, 1989. Address reprint requests to Dr C. J . Chapman, Department of Laboratory Services, Wellington Hospital, Private Bag, Wellington South, New Zealand.
0 1990 Wiley-Liss, Inc.
METHODS The metaphor this program uses is a simple one. The computer screen is considered to represent a window onto a large sheet of paper, upon which is drawn the family tree. This window can be moved around the family tree, or the entire pedigree can be viewed in miniature (Fig. 1).The program functions like a simple data base with a rather unusual method of displaying individual data items and the relationships between them. In concept, it functions in a manner similar to that of Hypercard on the Apple Macintosh. The program was originally written in the TDI implementation of Modula-2 [TDI Software, 19861, and includes within it the algorithm of the PREPLINK program from the LINKAGE package. Therefore, data can be output in the appropriate format for MLINK. However, this program was developed in order to demonstrate the feasibility and attractiveness of a graphic approach to pedigree data entry, and it was not intended that a complete pedigree analysis program be created. Part of the motivation behind the development of the current version was to make i t easier to calculate genetic risks in families, given disease status and genetic marker information. It has been used to calculate the likelihood of a series of Huntington disease families, using the G8 probe, and to calculate risks for possible carriers of the gene given this same information. [Stapleton, 19891. As with most window-based software, drop-down menus are used to alter default settings, open and close files, obtain help, and perform general housekeeping chores. Keyboard shortcuts are also provided for most commands, including the output options for MLINK. Four drop-down menus are provided (Fig. 2)-a HELP menu that has options giving advice on drawing pedigrees, using the mouse, and defining the factors that describe MLINK loci; a UTILITIES menu that controls the display of a listing of available genetic markers, the editing of marker information, the editing of disease locus data, a reduced view of the pedigree, spacing within the pedigree, and output for MLINK; a DEFAULTS menu that controls the on-screen information (pedigree numbers, surnames, forenames, and marker data); and a FILE menu t h a t has options providing for setting up a new family, opening family files, editing family information, printing the pedigree, saving the pedigree, redrawing the pedigree on the screen, abandoning the pedigree, and
Chapman
156
lisplay o f
-
Brown
-
family
Screen 0,fl
___II) ieee
Brown Alfred
e5ee
Smith Chris nB
22
/’
1 1
eeei
Brown Betty BC
8102
eie3
Smith
Brown
Joan AC
Dora AB
Smith E l s i e AC
12
11
12
b
9851 Smith
Bess AC 22
1 Fig. 1. Pedigree
2 Fig. 2. Drop-down menu
User Interfaces for Pedigree Analysis quitting the program. The mouse is used to move the screen around the pedigree, to select individuals for data editing and to move pedigree symbols. In order to demonstrate how the metaphor operates, the entry of a new family will be described briefly. The user selects a menu option “New Family,” and is presented with a dialogue box, into which is entered basic identifying data such as the family name and number, and the family’s genetic problem (Fig. 3). When this is completed, the screen is cleared, and the user specifies the position on the screen of the first member of the pedigree (perhaps the proband), by using the mouse as a pointer. This opens a dialogue for data entry of items such as pedigree number, name, genetic status, and marker data (Fig. 4).The program then draws on the screen the standard pedigree symbol appropriate to the individual’s sex and genetic status. All other relatives are specified by their relationship to an individual already on screen. The user clicks the mouse button on the symbol for that individual and selects the relationship of the new individual from the dialogue that pops up (Fig. 5). The program keeps track of the relationships between individuals, and draws connecting lines on screen as required. The process of adding to the pedigree is simple and intuitive, although there are some traps for the unwary, particularly in defining parents, as both parents must be defined by their relationship to their child. The program will present only valid options for selection, reducing the chances of errors in the entry of relationships. Information written on the screen can be any combination of pedigree number, first name, surname, maiden name, birth date, and marker data. Individuals are simply added to a stored pedigree by recalling the pedigree to the screen, locating the parents of the new individual, and defining a new child for them. If spacing is a problem, it can be altered by using one of the menu items. The program uses a structured presentation of marker data, the user selecting the chromosome on which the marker lies, and then entering the details of alleles and phenotypes, including the details of the factors as used by the LINKAGE package. Default factor definitions and phenotypes are generated by the program but are readily altered. The emphasis throughout the program is on a visual appearance that matches the habits of users-in this case, clinical geneticists. An important principle is that the user should not be involved in any effort coding data before it is entered into the computer.
157
6),that individual is defined as a proband, and the input file for MLINK is amended by the program to force risk calculation for that person.
DISCUSSION
Details of the use of the program will be supplied to inquirers, but suffice it to say that the emphasis throughout is on matching the operation of the program to the habits of its likely users. The visual appearance of data should, a s far as possible, match that already familiar to the user, and its manipulation should be as intuitive as the operating system will allow. Thus, users need to have a n elementary knowledge of program operation under GEM, but little other information should be necessary for them to attempt most of the simpler tasks the program requires. However, it is important that users understand the metaphor for data entry that the program uses. It must not be confused with that used by other programs, such as “Pedigree/Draw” for the Macintosh. The program is designed to input and display pedigree information. It has no facility to accept data from disk files or databases, although a simple method of doing so is under consideration. Regardless of this, there was no intention of producing a n IBM PC version of the “PedigreeiDraw” program. The data that the program needs in order to draw a pedigree is minimal-no more than one would need to draw a pedigree on a piece of paperand the form that marker data takes is entirely under the user’s control. Programmers undertaking program development with this type of graphics system have to undergo a considerable amount of re-education before they become proficient in the use of the graphic resources. The standard approach applied to most mainframe and PC software is not appropriate, and the principles behind the interaction with the user of the software have to be completely rethought. In fact, more programming time has to be spent on the user interface than on the details of data entry and manipulation. Anyone undertaking a full linkage analysis obviously requires much additional knowledge in order that they can avoid certain kinds of errors, and be able to understand the output from MLINK, but a t least the more egregious errors should be avoided by the suggested method of data entry. Data checking is simplified by the screen display of marker information and identifying information for individuals. Anyone who plans to use linkage analysis packages must have a clear underRESULTS standing of the genetic and statistical principles inFigure 6 shows a hypothetical pedigree of a family volved. Those who cannot look a t a simple pedigree and with Huntington disease, with the names, pedigree guess the range within which the linkage results are numbers, and marker information of the individuals likely to lie, probably should not be using such a package within it. To calculate the lod score for this family, the at all, but should have their analyses carried out by a n user simply selects the menu item “Output to MLINK,” experienced worker. They themselves will be unable to agrees with the program that all consanguinity loops note obvious errors in the analyses or data presentation. have been defined, and the program then generates the This program is presented for experienced analysts who locus information and the pedigree input files for MLINK. might wish to get the feel of a different approach to data How MLINK is run depends on the version used, and is entry, and is not intended for those who are uninitiated beyond the scope of this article. To calculate the genetic into the mysteries of linkage analysis. Novice users risk for a n individual, such as 0102 in the pedigree (Fig. would require a better validated program than this one,
Chapman
158
Family identifying information Family name :
-------------------
Family number :
----
I Show
family markers
I
pmmq
Genetic condition : ______________________________ To locate disease, enter first few letters and click on the button --> Find disease I
I
Mutation rate : 8,08,---, Phenotype kind ->
Chromosome :
--
(8inarlrl lAffectionl1 Quantitative
1
3 Fig. 3. Dialogue box screen with family identifying information.
Date o f birth : --/--/---Date of death : --/--I ---_ Sex (lMilFll?l]
I
fldopted
I
Genetic Status I I ? 1 I N 1 I A I IHemizugl I H e t e r o r w 1 1Homozuqj 1 Liability or trait class : -! ! ! Be VERY cautious if changing any of this group ! ! ! ID number : ------ Pedigree number : ---_-Mother's number : -----_ Father's number : ____-_ Spouse's number : _-__-_Second spouse : -----Third spouse : _-_--- Twin : ------
4 Fig. 4.Individual identifying information screen.
User Interfaces for Pedigree Analysis
Select a button t o identify the
I Daughter I I
Son
I I
type
Sister
of relative
I I Brother 1
Individuals o f unknown sex :-
IPartnerJ 1 -
pq
Twins :
5
I
Fig. 5. Relationship selection screen.
Desk File Defaults Utilities Help isplav of
- Brown -
Harkers Screen 0,8
family
pJi isee
100s
Brown Alfred
Brown
Betty
AA
BC 12
11
Dasoa
~MIB~
Smith
Smith
Chris AB
Joan
AC 12
22 1
a102 Brown Dora CIB
11
8103 Smith E l s i e AC 12
I
9051
Smith Bess
AC
22
6 Fig. 6. Risk-calculation screen.
D 4 S 1 0 Hn D 4 S 1 0 Ec
159
160
Chapman
not to mention attendance a t a formal course on linkage analysis. It is to be hoped that this paper will help to stimulate interest in graphic user interfaces for the programs used by geneticists. The designers of genetic data bases might also consider this approach, a s one of the fundamental items of information in such data bases is the relationships between persons in a pedigree. A graphic presentation of such information is a n integral part of data entry; its display is a logical extension of most clinical geneticists daily work habits. Copies of the IBM PC AT version of the program, on 51hinch high-density floppy disks, will be made available to any geneticist who may wish to explore this method of data entry and presentation. The Modula-2 source files and some illustrative pedigrees can be supplied along with the complied program, but a copy of GEMI3 will be required in order to run it. The IBM version of this program has several extensions that may be of interest. The pedigree drawings can be printed on any printer for which GEM has a driver, and haplotypes
can be shown graphically as bar codes. These bar codes can be used to illustrate to counsellees how haplotypes have been transmitted within their family. We now use the program to document all our clinic families.
ACKNOWLEDGMENTS I a m grateful to the two anonymous reviewers whose suggestions have substantially improved the text of this paper. Dr. R. J. M. Gardner has provided enthusiastic support for this project.
REFERENCES Lathrop GM, Lalouel JM (1984): Easy calculations of lad scores and genetic risks on small computers. Am J Hum Genet 36:460-465. Ott J (1974): Estimation of the recombination fraction in human pedigrees: Efficient computation of the likelihood for human linkage studies. Am J Hum Genet 26:588-597. Stapleton PM (1989): Molecular genetics of restriction fragment length polymorphisms linked to the Huntington disease locus. Ph.D. thesis, University of Auckland, Auckland, New Zealand. TDI Modula-2iST (1986): TDI Software Inc. Dallas.
Occasional Paper A Visual Interface to Computer Programs for Linkage Analysis Cyril J. Chapman Department of Laboratory Seruices, Wellington Hospital, Wellington, New Zealand
This paper describes a visual approach to the input of information about human families into computer data bases, making use of the GEM graphic interface on the Atari ST. Similar approaches could be used on the Apple Macintosh or on the IBM PC AT (to which it has been transferred). For occasional users of pedigree analysis programs, this approach has considerable advantages in ease of use and accessibility. An example of such use might be the analysis of risk in families with Huntington disease using linked RFLPs. However, graphic interfaces do make much greater demands on the programmers of these systems. KEY WORDS: user interface, pedigree analysis, linkage analysis
INTRODUCTION Any clinical geneticist who has attempted to analyze a human pedigree for genetic risks using either LIPED [ o t t , 19741 or the LINKAGE package [Lathrop and Lalouel, 19841, will be aware of the considerable difficulties experienced by anyone who uses these programs only occasionally. The methods used for data entry in these programs are not intuitive for a clinical geneticist, and data verification after entry into the computer files is difficult. Anyone who has access to a microcomputer system that uses a graphics-based interface with windows, menus, icons, and a pointer (mouse) will know that such difficulties are not a necessary part of computing. This paper describes a program that uses the graphic interface provided by GEM on the Atari 520ST and the IBM PC AT. A similar data-entry system could be developed on a n Apple Macintosh or any of the other graphical interfaces now available, such as the OSi2 Presentation Manager or Microsoft Windows. Received for publication November 4, 1988; revision received October 11, 1989. Address reprint requests to Dr C. J . Chapman, Department of Laboratory Services, Wellington Hospital, Private Bag, Wellington South, New Zealand.
0 1990 Wiley-Liss, Inc.
METHODS The metaphor this program uses is a simple one. The computer screen is considered to represent a window onto a large sheet of paper, upon which is drawn the family tree. This window can be moved around the family tree, or the entire pedigree can be viewed in miniature (Fig. 1).The program functions like a simple data base with a rather unusual method of displaying individual data items and the relationships between them. In concept, it functions in a manner similar to that of Hypercard on the Apple Macintosh. The program was originally written in the TDI implementation of Modula-2 [TDI Software, 19861, and includes within it the algorithm of the PREPLINK program from the LINKAGE package. Therefore, data can be output in the appropriate format for MLINK. However, this program was developed in order to demonstrate the feasibility and attractiveness of a graphic approach to pedigree data entry, and it was not intended that a complete pedigree analysis program be created. Part of the motivation behind the development of the current version was to make i t easier to calculate genetic risks in families, given disease status and genetic marker information. It has been used to calculate the likelihood of a series of Huntington disease families, using the G8 probe, and to calculate risks for possible carriers of the gene given this same information. [Stapleton, 19891. As with most window-based software, drop-down menus are used to alter default settings, open and close files, obtain help, and perform general housekeeping chores. Keyboard shortcuts are also provided for most commands, including the output options for MLINK. Four drop-down menus are provided (Fig. 2)-a HELP menu that has options giving advice on drawing pedigrees, using the mouse, and defining the factors that describe MLINK loci; a UTILITIES menu that controls the display of a listing of available genetic markers, the editing of marker information, the editing of disease locus data, a reduced view of the pedigree, spacing within the pedigree, and output for MLINK; a DEFAULTS menu that controls the on-screen information (pedigree numbers, surnames, forenames, and marker data); and a FILE menu t h a t has options providing for setting up a new family, opening family files, editing family information, printing the pedigree, saving the pedigree, redrawing the pedigree on the screen, abandoning the pedigree, and
Chapman
156
lisplay o f
-
Brown
-
family
Screen 0,fl
___II) ieee
Brown Alfred
e5ee
Smith Chris nB
22
/’
1 1
eeei
Brown Betty BC
8102
eie3
Smith
Brown
Joan AC
Dora AB
Smith E l s i e AC
12
11
12
b
9851 Smith
Bess AC 22
1 Fig. 1. Pedigree
2 Fig. 2. Drop-down menu
User Interfaces for Pedigree Analysis quitting the program. The mouse is used to move the screen around the pedigree, to select individuals for data editing and to move pedigree symbols. In order to demonstrate how the metaphor operates, the entry of a new family will be described briefly. The user selects a menu option “New Family,” and is presented with a dialogue box, into which is entered basic identifying data such as the family name and number, and the family’s genetic problem (Fig. 3). When this is completed, the screen is cleared, and the user specifies the position on the screen of the first member of the pedigree (perhaps the proband), by using the mouse as a pointer. This opens a dialogue for data entry of items such as pedigree number, name, genetic status, and marker data (Fig. 4).The program then draws on the screen the standard pedigree symbol appropriate to the individual’s sex and genetic status. All other relatives are specified by their relationship to an individual already on screen. The user clicks the mouse button on the symbol for that individual and selects the relationship of the new individual from the dialogue that pops up (Fig. 5). The program keeps track of the relationships between individuals, and draws connecting lines on screen as required. The process of adding to the pedigree is simple and intuitive, although there are some traps for the unwary, particularly in defining parents, as both parents must be defined by their relationship to their child. The program will present only valid options for selection, reducing the chances of errors in the entry of relationships. Information written on the screen can be any combination of pedigree number, first name, surname, maiden name, birth date, and marker data. Individuals are simply added to a stored pedigree by recalling the pedigree to the screen, locating the parents of the new individual, and defining a new child for them. If spacing is a problem, it can be altered by using one of the menu items. The program uses a structured presentation of marker data, the user selecting the chromosome on which the marker lies, and then entering the details of alleles and phenotypes, including the details of the factors as used by the LINKAGE package. Default factor definitions and phenotypes are generated by the program but are readily altered. The emphasis throughout the program is on a visual appearance that matches the habits of users-in this case, clinical geneticists. An important principle is that the user should not be involved in any effort coding data before it is entered into the computer.
157
6),that individual is defined as a proband, and the input file for MLINK is amended by the program to force risk calculation for that person.
DISCUSSION
Details of the use of the program will be supplied to inquirers, but suffice it to say that the emphasis throughout is on matching the operation of the program to the habits of its likely users. The visual appearance of data should, a s far as possible, match that already familiar to the user, and its manipulation should be as intuitive as the operating system will allow. Thus, users need to have a n elementary knowledge of program operation under GEM, but little other information should be necessary for them to attempt most of the simpler tasks the program requires. However, it is important that users understand the metaphor for data entry that the program uses. It must not be confused with that used by other programs, such as “Pedigree/Draw” for the Macintosh. The program is designed to input and display pedigree information. It has no facility to accept data from disk files or databases, although a simple method of doing so is under consideration. Regardless of this, there was no intention of producing a n IBM PC version of the “PedigreeiDraw” program. The data that the program needs in order to draw a pedigree is minimal-no more than one would need to draw a pedigree on a piece of paperand the form that marker data takes is entirely under the user’s control. Programmers undertaking program development with this type of graphics system have to undergo a considerable amount of re-education before they become proficient in the use of the graphic resources. The standard approach applied to most mainframe and PC software is not appropriate, and the principles behind the interaction with the user of the software have to be completely rethought. In fact, more programming time has to be spent on the user interface than on the details of data entry and manipulation. Anyone undertaking a full linkage analysis obviously requires much additional knowledge in order that they can avoid certain kinds of errors, and be able to understand the output from MLINK, but a t least the more egregious errors should be avoided by the suggested method of data entry. Data checking is simplified by the screen display of marker information and identifying information for individuals. Anyone who plans to use linkage analysis packages must have a clear underRESULTS standing of the genetic and statistical principles inFigure 6 shows a hypothetical pedigree of a family volved. Those who cannot look a t a simple pedigree and with Huntington disease, with the names, pedigree guess the range within which the linkage results are numbers, and marker information of the individuals likely to lie, probably should not be using such a package within it. To calculate the lod score for this family, the at all, but should have their analyses carried out by a n user simply selects the menu item “Output to MLINK,” experienced worker. They themselves will be unable to agrees with the program that all consanguinity loops note obvious errors in the analyses or data presentation. have been defined, and the program then generates the This program is presented for experienced analysts who locus information and the pedigree input files for MLINK. might wish to get the feel of a different approach to data How MLINK is run depends on the version used, and is entry, and is not intended for those who are uninitiated beyond the scope of this article. To calculate the genetic into the mysteries of linkage analysis. Novice users risk for a n individual, such as 0102 in the pedigree (Fig. would require a better validated program than this one,
Chapman
158
Family identifying information Family name :
-------------------
Family number :
----
I Show
family markers
I
pmmq
Genetic condition : ______________________________ To locate disease, enter first few letters and click on the button --> Find disease I
I
Mutation rate : 8,08,---, Phenotype kind ->
Chromosome :
--
(8inarlrl lAffectionl1 Quantitative
1
3 Fig. 3. Dialogue box screen with family identifying information.
Date o f birth : --/--/---Date of death : --/--I ---_ Sex (lMilFll?l]
I
fldopted
I
Genetic Status I I ? 1 I N 1 I A I IHemizugl I H e t e r o r w 1 1Homozuqj 1 Liability or trait class : -! ! ! Be VERY cautious if changing any of this group ! ! ! ID number : ------ Pedigree number : ---_-Mother's number : -----_ Father's number : ____-_ Spouse's number : _-__-_Second spouse : -----Third spouse : _-_--- Twin : ------
4 Fig. 4.Individual identifying information screen.
User Interfaces for Pedigree Analysis
Select a button t o identify the
I Daughter I I
Son
I I
type
Sister
of relative
I I Brother 1
Individuals o f unknown sex :-
IPartnerJ 1 -
pq
Twins :
5
I
Fig. 5. Relationship selection screen.
Desk File Defaults Utilities Help isplav of
- Brown -
Harkers Screen 0,8
family
pJi isee
100s
Brown Alfred
Brown
Betty
AA
BC 12
11
Dasoa
~MIB~
Smith
Smith
Chris AB
Joan
AC 12
22 1
a102 Brown Dora CIB
11
8103 Smith E l s i e AC 12
I
9051
Smith Bess
AC
22
6 Fig. 6. Risk-calculation screen.
D 4 S 1 0 Hn D 4 S 1 0 Ec
159
160
Chapman
not to mention attendance a t a formal course on linkage analysis. It is to be hoped that this paper will help to stimulate interest in graphic user interfaces for the programs used by geneticists. The designers of genetic data bases might also consider this approach, a s one of the fundamental items of information in such data bases is the relationships between persons in a pedigree. A graphic presentation of such information is a n integral part of data entry; its display is a logical extension of most clinical geneticists daily work habits. Copies of the IBM PC AT version of the program, on 51hinch high-density floppy disks, will be made available to any geneticist who may wish to explore this method of data entry and presentation. The Modula-2 source files and some illustrative pedigrees can be supplied along with the complied program, but a copy of GEMI3 will be required in order to run it. The IBM version of this program has several extensions that may be of interest. The pedigree drawings can be printed on any printer for which GEM has a driver, and haplotypes
can be shown graphically as bar codes. These bar codes can be used to illustrate to counsellees how haplotypes have been transmitted within their family. We now use the program to document all our clinic families.
ACKNOWLEDGMENTS I a m grateful to the two anonymous reviewers whose suggestions have substantially improved the text of this paper. Dr. R. J. M. Gardner has provided enthusiastic support for this project.
REFERENCES Lathrop GM, Lalouel JM (1984): Easy calculations of lad scores and genetic risks on small computers. Am J Hum Genet 36:460-465. Ott J (1974): Estimation of the recombination fraction in human pedigrees: Efficient computation of the likelihood for human linkage studies. Am J Hum Genet 26:588-597. Stapleton PM (1989): Molecular genetics of restriction fragment length polymorphisms linked to the Huntington disease locus. Ph.D. thesis, University of Auckland, Auckland, New Zealand. TDI Modula-2iST (1986): TDI Software Inc. Dallas.
