GENFILES A computerized medical genetics information network. III. CHROMO The cytogenetics databaseкод для вставкиСкачать
American Journal of Medical Genetics 7:267-278 (1980) GENFILES A Computerized Medical Genetics Information Network. 111. CHROMO: The Cytogenetics Database William D. Loughman, Douglas C. Mosher, and Charles J. Epstein Department of Pediatrics, and the Computer Center, University of California, San Francisco The computer database CHROMO is the cytogenetic branch of the GENFILES medical genetics information system. Complete cytogenetic laboratory data are maintained in a format that allows detailed searches of client records. Both the standard and extended Paris nomenclature are used. Key words: cytogenetics, medical genetics, CHROMO, information storage and retrieval, database systems, CENFILES INTRODUCTION Karyotype analyses, even from normal persons, can produce very complex data. A complete analysis should contain some notation both of normal variation and of those atypical findings with as yet unknown significance. It is generally quite cumbersome to sift such information-rich records for evidence of a time-dependent incidence of abnormalities or for correlations with phenotypes. Addditionally, manual searching and tabulation are prone to error. Therefore, the computer database named CHROMO has been designed to perform storage, search, and correlation tasks. All of the cytogenetic data obtained in a large prenatal diagnosis program [Golbus et al, 19791 are stored in CHROMO, as well as the analyses performed for a large genetic counseling clinic. CHROMO now contains the cytogenetic results from over 5,500 prenatal analyses and over 2,600 clinical analyses. DATABASE DESCRIPTION CHROMO is a hierarchical archive file with 26 data fields containing both karyotype descriptions and related laboratory data. About half of the fields are used for patient identification, clinic information, laboratory quality control, or database interconnection. The remaining fields are devoted to a segmented description of the patient’s karyotype, along with any comments or information not accommodated in a specific data Received for publication April 21, 1980; revision received July 30, 1980. Address reprint requests to Dr. William D. Loughman, Cytogenetics Laboratory, AC-28, University of California, San Francisco, CA 94143. 0148-7299/80/0703-0267$02.000 1980 Alan R. Liss, Inc. Loughman, Mosher, and Epstein 268 field. Paris cytogenetic nomenclature is fully utilized in the stored karyotype descriptions. The new analyses that may require updating or correction are stored temporarily in an identically structured input file, the content of which is peridocally incorporated into CHROMO. Table I shows the data structure of CHROMO, with its field names and a short description of the type of data each contains. In this description, LEVEL is a hierarchical branch-point where multiple entries may be appended to existing data. Fields may be identified for data entry or reports by number (not shown), fieldname, or synonym. TYPE and LENGTH dictate the format of a field’s content. For example, the alphabetic field SURNAME may contain up to 16 characters of any kind. The BIRTHDATE field, however, contains only six integers, arranged to provide two digits each for the year, month, and day (YYMMDD). The definitions for the 26 fields in CHROMO shown in Table I are those stored within the computer, but a fuller explanation of certain fields follows. RF’RT is the date that a Birth Defects Center physician examined the laboratory results. STORAGE is a room number in the University, showing where the medical record is kept. INDICATION is the reason the study was performed. This is a field with a closely controlled but expandable vocabulary. SOURCE and PHYSICIAN help keep track of where reports TABLE 1. Brief Description of File CHROMO LEVEL FIELDNAME SYNONYM 1 2 LAST FIRST BIRTHDATE FAMNUM IN HARVEST OUT ENTER LOC WHY CLINIC MD LABN RAMPASS LINE CNUM SEXC SIGN TRIPLET (:SOME1 CSOME2 PARENT BAND1 BAND2lSTAIN COM QD 3 4 5 SURNAME GIVEN DOB FAMILY INTAKE HRVST RPRT FILED STORAGE INDICATION SOURCE PHYSICIAN LAB NUM RAMPASS M CN sx 6 7 S TRI c1 C2 PRNT B1 B2S COMMENT BRIEFLY TYPE A A I A I I LENGTH 16 14 6 8 A 6 6 6 6 10 18 12 16 10 8 I 2 3 I 6 4 4 A A A A A 4 6 42 12 1 I A A A A A A A I A A A A 5 DEFINITION FAMILY NAME O F SUBJECT FIRST NAME OF SUBJECT BIRTHDATE OF SUBJECT: YYMMDD KEY FOR LINKING FAMILY MEMBERS DATE O F SAMPLE RECEIPT: YYMMDD DATE O F CULTURE HARVEST: YYMMDD DATE MD CERTIFIED RESULTS DATE RECORD ENTERED TO COMPUTER STORAGE LOCATION OF LAB RECORD REASON FOR KARYOTYPE ORIGINATING CLINIC REFERRING PHYSICIAN AC28 ACCESSION NUMBER . . . A SECURITY DEVICE CELL LINE NUMBER IF MOSAIC KARYOTYPE NUMBER O F CHROMOSOMES IN KARYOTYPE SEX CHROMOSOMES. QUOTE: SEE TO RIGHT PLUS/MINUS SPECIFIED CHROMOSOME “PARIS CONFERENCE” TRIPLET CODE IST (OR ONLY) CHROMOSOME 1NVOLVF.D 2ND (IF ANY) CHROMOSOME INVOLVED CONTRIBUTOR PARENT (OR) NUMBER ALIKE AFFECTED BAND OF C1 AFFECTED BAND O F C2 (OR) STAIN CODF ADDITIONAL INFORMATION SIGNIFICANT FEATURE OF KARYOTYPE GENFILES: The Cytogenetics Database 269 must be sent. LAB NUM is the number assigned to each sample entering the laboratory. Every sample gets a unique number consisting of a letter (A, B, G, etc) designating the type of sample (amniotic fluid, blood, gonad, etc), and a number that is sequential within years. For example: B80201 is the 201st blood sample in 1980. RAMPASS is a field whose contents, normally blank, become an additional password. M or LINE is the number of the cell line in a mosaic patient. This may be 1 , 2 , 3 , etc. If 0, no mosaicism was detected. COMMENT allows entry of anything not covered in the record format. Usually, ratios of abnormal count to total count (in an analysis), stains used, and similar information are stored here. The remaining fields are a part of the karyotype itself. KARYOTYPE REPRESENTATION IN CHROMO The karyotype begins at level 5, with field “M” (LINE). In order to see how the data are structured within CHROMO, consider the following karyotype: mos 46,XX,-S,+der(5)t(5;16) (q35;q23)pat,var(16)(qll ,CBG43)mat,var(l3) (s,GTG44)*2/46,XX, var(l6) (41 1,CBG43)mat, var(13)(s,CTG44)*2 Ignoring the chromosome number and sex designation, it is possible to consider all elements between commas (except within parentheses) as karyotpe units that refer uniquely to a single structure. Inspection shows that each such unit, in principle, contains five major parts. These are a sign (+/-), a leading triplet code, two sets of parenthetical expressions, and a trailing qualifier. If anything to the right of the sex designation exists, then at least some of the content of the first set of parentheses exists. This content always is preceded by either or both of a sign and a triplet. The parenthetical expressions contain at least two elements each, but in degnerate cases one or more elements may be blank, as may be the trailing qualifier. The content of the first parenthetical expression always refers to at least one chromosome. That of the second always refers either to at least one breakpoint or to a band number and a code describing how the band behaved under a particular stain. If the derivative-5 chromosome in the example were to be written in the conventional way, but incorporating these concepts, it would look like this: +der(5 ;. . .)(. . .;. . .) . . . . . .t(5 ;16)(q35 ;q23)pat In this representation, an ellipsis is a place holder indicating an unused element. This is the karyotype representation used in CHROMO, producing the configuration shown in Table 11. The distinct subdivisions are arranged one above the other. Corresponding parts are aligned vertically in an orderly repetition of fields TRI, C1, C2, B1, B2S, and PARENT. Visually, the arrangement suggests a matrix, and it might be treated as such. But CHROMO is a hierarchically structured file that uses three levels for the karyotype. Thus, in CHROMO the matrix does not exist. The representation is useful though, and it is retained in the data input form used by the Cytogenetics Laboratory (Fig. 1). It is possible to descibe almost every chromosome abnormality in this format, even those involving more than two chromosomes and more than two breakpoints. For example, this hypothetical fourchromosome translocation: 270 Loughman, Mosher, and Epstein Fig. 1. Form used for recording input data for CHROMO. The regions contained within the heavy lines are filled o u t by the referring professional. t(l;2;3;4)(q 1 1;q 12;ql3,14) becomes, for entry to CHROMO: TRI C1 C2 B1 B2S t (1 ;2) (qll ( ;3) ( ;q13) ( ;4) ( ;c114) The one-line karyotype becomes three branches of a hierarchical "tree," uniquely tied to each other by at least one common node. In this case, the node is the SIGN field, on which the t h e e records depend. The absence of C1 for the records pertaining to chromosomes 3 and 4 is a convention used in CHROMO to indicate multichromosome aberrations. If there were more than one such aberration in a karyotype, and possible ambiguity as to which chromosome was in which translocation, a qualifier would be added. GENFILES: The Cytogenetics Database 271 Some very complex karyotypes might resist unambiguous description. These would be “flagged” in the TRI field, and the karyotype would be written conventionally in the COMMENT field. We have not encountered a need to do this, although CHROMO has over 11,000 entries. There are some problems with the Paris nomenclature itself. For example, the standard designation for a Robertsonian translocation may not make clear which chromosome arms are actually present in the karyotype. This and related ambiguities are resolved in CHROMO by using the field name BRIEFLY. In practice, the most common application of this field has been in rapid searching for broad categories of chromosome aberration - eg, 5 q t , TRI21, and the like. REPORTING FROM THE DATABASE Two related methods are available to obtain reports from CHROMO. The most general is to use the standard RAMIS report language, typing a series of instructions at the terminal, l i e by line. The easier method is to use GENFILES preprogrammed procedures. The list of these has become quite large. Included are commands that provide complete karyotypes and other data over a range of group selection criteria, compute and tabulate duration of culture for selected sample types, and graph the frequencies of occurence of a field’s content. Most routines are quite general and require the addition of a few parameters. If the parameters are neglected, the computer prompts for their entry. In many cases the parameters permit restriction of a search range or permit conditional search. Three examples of GENFILES routines that address CHROMO follow. An example of a special Cytogenetics Laboratory summary is shown in Table 111. The report in Table 111 contains results of amniotic fluid cell analyses sequenced by laboratory number. Amniotic cell karyotype A5025, from a 38-year-old woman, is of a male fetus with variants of chromosomes 13,15, 19, and 21; an inversion of chromosome 9 between p12 and q12; and a reciprocal balanced translocation between chromosomes 1 8 and 21. Ariother type of report provides information about the technical aspects of the laboratory procedures. A summary of the duration in culture of amniotic fluid cells calculated from the intake dates and the harvest dates for all months in 1979 is shown in Table IV. Minimum, average, and maximum values are displayed, and “no growth” results are ignored in order to show true minima. Procedures are also available for extracting various types of demographic data from CHROMO. An example of this is the table and graph of the number of mothers in yearly age categories of fetuses prenatally diagnosed as having simple non-mosaic trisomy 21 shown in Figure 2. The women’s ages were calculated by subtracting birthdates from the date of amniocentesis. The examples given above are straightforward and uncomplicated uses of GENFILES. Much more complex search and reporting methods are possible. One involves the concept of virtual files. These are not data files in the usual sense. Rather, they are empty files that contain instructions for obtaining real data from one or more real files. They can be used for many quite different purposes. For example, a single low-level selector (ie, BRIEFLY) can be used to obtain the complete karyotype, which is the whole hierarchical patient record, without an expensive file inversion. Results of such a search to obtain from CHROMO complete karyotypes for all patients having “partial trisomy of 5q” are shown in Table V. 272 Loughman, Mosher, and Epstein TABLE 11. A Karyotype Arranged for Entry to CHROMO M CN 1 46 sx xx S - + 2 46 xx TRI der t var var var var C1 5 5 5 16 13 16 13 C2 B1 B2S PRNT COM BRIEFLY 12/30 mon5q part. tril6q part. 18/30 5q+ mos Pat 16 q35 q23 q l l CBG43 GTC44 s CBG43 s CTG44 qll mat *2 mat *2 AGE COUNT --- ----- 20 35 36 37 39 40 41 42 43 48 AGE 1 6 5 5 10 3 5 2 4 1 COUNT C 20 22 24 26 26 30 32 34 36 36 40 42 44 46 48 C C C C FROM DATA IN COMPUTER ON ( 1 4 / 0 7 / 8 0 AT 6.05.21 Fig. 2. Ages of women found to have fetuses with trisomy 21 at time of receipt of amniotic fluid sample in laboratory. The age frequency above age 15 years is tabulated and plotted. AS026 JE AS032 YO 38 46XY 1 VAR VAR VAR VAR VAR 19 VAR S 21 H S Q12 PI1 012 012 PI3 PI2 S Q12 21 GTC43 CBC23 B2S GTG43 GTGO2 GTG43 CBG43 GTG43 GTG02 412 CTGO2 GTGO2 GTC43 Q12 P I 3 CTG43 P I 3 GTG43 P116 Q112 PI3 H B1 14 19 22 9 INV 19 19 21 9 VAR VAR INV VAR 1s 18 C2 PRNT UCSF-PRENATL MSG UCSF-PRENATL MSC PHYSICIAN GTG/CBG/RHG GTG/CBG/RHG RHG1CBG:DlTTO CTG/CBG/RHG G r c ONLY SEE ~ 2 2 9 3 GTG/CBC/RIIG GTG/CBG/RHG UCSF-PRENATL MSG UCSF-PRENATL MSG UCSF-PRENATL MSG UCSF-PRENATL MSG UCSF-PRENATL MSC UCSF-PRENATL MSG UCSF-PRENATL MSG UCSF-PRENATL MSG 33/33 CELLS (2 FLASKS) GTC/RHC/NOR/QFQ/CBG ON PREPONDERANCE OF EVIDENCE CTGIRHG SEE A2748 GTG:DITTO COMMENT SOURCE *Names in the report have been coded by the computer for publication. The normal report contains the patients’ full names. LU AS027 MO DO A5028 NO PH AS029 RE MA AS030 BE CO AS031 LA 13 VAR VA R RCP 36 40XX 36 46XX 38 46XX 37 46XY 35 46XY 36 46XY 21 VAR EL GO 9 C1 VAR 3s 2 41XY” 40 46XY 38 46XY PA AS023 DA YO AS024 FR KA AS025 CO M CNSX S TRI LABNUM LASTNAME FlRSTNAME INDICATION UNIVERSITY O F CALIFORNIA, SAN FRANCISCO CYTOGENETICS LABORATORY, AC-28 (415) 666-3121 CYTOCENETIC ANALYSES REPORTED BY FIELD LABN FROM AS000 TO A6000. FILE: NEWSTUFF 02/28/80 15.48.27 TABLE 111. Portion of Weekly Cytogenetia Laboratory Report of Amniotic Fluid Cell Karyotypes* INV9 lNV9 T(18;21) BAL BRIEFLY 274 Loughman, Mosher, and Epstein TABLE IV. Average, Minimum, and Maximum Times in Days Required to Culture Amniotic Fluid Cells for the Cited Year (1979) and Months* YR MN AVE TH MIN TH MAX TH 79 1 2 3 4 5 6 7 8 9 10 11 12 15.65 14.42 14.15 15.69 15.54 16.27 16.43 17.55 16.05 15.44 14.98 15.09 9.00 10.00 10.00 10.00 11.00 12.00 10.00 7.00 10.00 11.00 11.00 10.00 43.00 20.00 28.00 24.00 27.00 49.00 50.00 43.00 25.00 21.00 20.00 21.00 *“No-growths” are ignored in order t o show true minima. TABLE V. Part of a Report of A Search of CHROMO for all Patients With Partial Trisomy of 5q VIRKARYO RAMEXEC REPORTED ON FIELD BRIEFLY=TRISQ-PART. FROM DATA IN COMPUTER ON 04/07/80 AT 10.46.20 SURNAME GIVEN DOB c1 C2 5 9 16 16 16 16 RCP 5 9 16 16 5 T +DER - 5 16 16 16 LABNUM M CNSX STRI DE s799 FE 790509 T VAR +DER 46XY - SA 46XY A4 184 550119 T VAR +DER - H76204 DO 876301 BL DE 876309 RI B78350 46XX 0 46XY 0 46XX DA 46XY T +DER 7 10320 T VAR +DER - 5 16 16 2 13 15 19 22 2 2 PKNT B1 b25 Q34 H Q24 GTG23 TRI5q-PART. MONl6Q-PART. MAT MAT? MAT 16 MAT 434 H Q24 RHG23 Q34 ~ 2 4 Q34 Q24 Q34 Q 24 TRISQ-PART. MON 16Q-PART. MAT 5? P24 S S Q12 PI 2 NEW TRISQ-PART. MON16Q-PART. T (5;16) BAL TR15Q-PART. MON 16Q-PART. MAT 16 BRlEFLY 4 31 CTC43 CTG43 GTG 32 GTG43 T (2;5?) TRISQ-PART. MON2P-PART. CENFILES: The Cytogenetics Database 275 PASSWORDS, CON F IDENTI ALI TY, A N D DATA SECU R IT Y Although standarized procedures are used for data transfers, mistakes can compromise the integrity of the database. Consequently, such maneuvers are performed only by a small number of well-trained personnel who have been provided with the proper password. A second password is used to allow staff members (clinicians, counselors) the privilege of searching the database and reporting from it, without the capability to make changes. The confidential nature of the data is assured through two devices. The “secure” version for CHROMO is called CHROMOS. Patient identification fields are not available to persons who access CHROMOS, no matter what password is issued. Attempts to obtain the patient identifiers cause complete failure of the report request. The second device assuring confidentiality is the RAMPASS field shown in the fde description of CHROMO (Table I). For most patients records this field has no content; it is blank. If an entry exists, it becomes a password which can block access to all data below it in the file structure, namely the complete karyotype. Typically, this method is used to protect employee’s test results from inspection by co-workers, but it has applications as well for cooperative databases. DISCUSS ION There are only a handful of computerized information systems the major emphasis of which is the handling of cytogenetic data. Most of these employ the older computer technology current when they were designed [Human Cytogenetic Registers, 19771. To emphasize the use of a modern computer database management system (DBMS) and the flexibility and ease of use that can be obtained through use of a DBMS as a cytogenetic register, we shall compare CHROMO with only one of these existing systems, the Interregional Cytogenetic Register System (ICRS) [Prescott et al, 19781. The ICRS stores chromosome data in a well-defined matrix, each row of which records two chromosomes and corresponding band numbers or breakpoints. There is no limit on the number of rows that might be used to code a complete karyotype. An “Aberration Code” in each row records the general class of each aberration. Generally, as is the case for sex chromosome descriptions, this is a fwed code with limited possibilities for expansion. It would appear to be cumbersome to code and then retrieve the more complex chromosome aberrations involving three or more chromosomes, or to provide unambigous descriptions of multiple aberrations within the same chromosome. When these occur, the lCRS stores complete karyotypes in a subsidiary field. Direct search for subunits within this format would be awkward. The approach in CHROMO is generally similar, but the d e f i e d matrix is replaced by a hierarchical data structure. Instead of one chromosome aberration per row, as in the ICRS matrix, CHROMO stores each element of an aberration in a “row,” with no limit t o the number of rows per aberration. This, in addition to easily defined subfields for any entry, affords great specificity for data searching. The “Aberration Code” used in the ICRS is most closely represented in CHROMO by the field BRIEFLY. But BRIEFLY contains a standard description of the effect, not the form, of each element in the chromosome aberration. Data entries are not coded in CHROMO, except insofar as the standard cytogenetic nomenclature is a code. Insomuch as the standard nomenclature is hugely expansible, so also are the “standard” entries to CHROMO. There is no obvious limit to the 276 Loughman, Mosher, and Epstein growth of the database lexicon. There is also no need to refer to code equivalences. Knowledge of cytogenetic nomenclature generally is sufficient, and recourse to written-out karyotypes has not been necessary. The batch processing ICRS system is written in COBOL, and it is therefore relatively difficult to alter either fde formats or procedures used to report from files. CHROMO is interactive with an on-line database. It is very easy to reprogram the system, at least for obtaining reports, and this is done often by genetics staff. An on-line computer system will be more expensive, at least in memory costs, than a system utilizing magnetic tape or card deck storage of data. CHROMO is no exception to this principle, and as it grows some aged data undoubtedly will be stored off-line. FUTURE PLANS The RAMIS database management system was designed for use in a multiple subsidiary large corporation environme,it, and these design provisions influence the potential for growth and adaptation of CHROMO. Like the other GENFILES subsystems, CHROMO should be very easy to use as part of a distributed information network. This concept is implemented at present using a laboratory affiliated with our own institution. Our experience to date has provided no serious problems of a technical nature. We plan expansion to four additional laboratories in a trial of the distributed network structure. As a necessary part of this expansion, CHROMO will be modified to enhance further the strict confidentiality of patient records. One step will be division of CHROMO into two parts, isolating patient identification data into separate files under the sole control of the individual cooperating laboratories. The remaining large archive file, with no patient identification at all, will be accessible to all laboratories in the network. The two parts are easily linked, permitting the retrieval of complete patient records only by the laboratory possessing the necessary linkage information. Management of different laboratories’ conflicting needs and goals, and file adaptations to meet these conflicts, are expected to be important problems. These growing pains are expected to include problems common to all distributed or cooperative computer ventures [Bernard, 19791. ACKNOWLEDGMENTS The authors wish to acknowledge the assistance of Ms Karen Michalski for aid in the production of figures and tables from CHROMO, and Ms Mary Joan Fredette for typing the manuscripts. This work was supported in part by PHS grant T32GM07085-03, by a grant from The National Foundation - March of Dimes, and by a contract with the California State Council on Developmental Disabilities. Dr. Epstein is an Investigator of the Howard Hughes Medical Institute. REFERENCES Bernard D (1979): Management problems in cooperative computing. Computing Surveys 11: 3-17. Center for Disease Control (November-December 1973, Issued March 1974): Congenital Malformation Surveillance. Cole J, Conneally PM, Hodes ME, Merritt AD (1978): Genetic family history questionnaire. J Med Genet 15:lO-19. Conneally PM (1975): Genetic data base. J CIin Comp 4:231-233. GENFILES: The Cytogenetics Database 277 Emery AEH (1976a): A computerized “at risk” register for genetic disease. In Turnbull AC, Woodford FP (eds): “Prevention of Handicap Through Antenatal Care.” Amsterdam: Elsevier, pp 13- 19. Emery AEH (1976b): RAPID: A genetic register system for the ascertainment and prevention of inherited disease. In Emery AEH, Miller JE (eds): “Registers for the Prevention and Detection of Genetic Disease,” New York: Symposia Specialists, p p 53-63. Emery AEH, Brough C, Crawfurd M, Harper P, Harris R, Oakshott G (1978): A report on genetic registers. J Med Genet 15:435-442. Emery AEH, Elliott D, Moores HM, Smith C (1974): A genetic register system (RAPID). J Med Genet 1 1 :145-15 1. Epstein CJ, Erickson RP, Hall BD, Golbus MS (1975): The center-satellite system for the widescale distribution of genetic counseling services. Am J Hum Genet 27:322-332. Epstein CJ (1979): Evolution of a center-satellite system for genetic counseling. In Porter IH, Hook EB (eds): “Service and Education in Medical Genetics” New York: Academic Press, pp 151-161. Golbus MS, Loughman WD, Epstein CJ, Halbasch G, Stephens JD, Hall BD (1979): Prenatal genetic diagnosis in 3000 aminocenteses. N Engl J Med 300:157-163. Hook EB (1976): Genetic registers: Types, goals, methods, and limitations. In Emery AEH, Miller JR (eds): “Registers for the Prevention and Detxxtion of Genetic Disease.” New York: Symposia Specialists, pp 9-29. Human Cytogenetic Registers (1977): A description of nine systems together with some recommendation. New York: Birth Defects: Orig Art Ser 13(4). The National Foundation-March of Dimes. Jacobs SE, Meyers DA, Klein SJ, Merritt AD (1977): A system for the retrieval of family structured information. Comput Biol Med 7:281-290. Loughman WD, Mosher DC (1 978): A new cytogenetics/medical genetics computerized data management system. Am J Hum Genet 30:87a. Loughman WD, Mitchell JA, Mosher DC, Epstein CJ (1980a): GENFILES: A computerized medical genetics information network. 1. An overview. Am J Med Genet (this volume). Loughman WD, Mosher DC, Epstein CJ (1980b): GENFILES: A computerized medical information network. 111. CHROMO: The cytogenetics database. Am J Med Genet (thisvolume). McKusick VA (1978): Mendelian Inheritance in Man.” Ed 5. Baltimore: Johns Hopkins University Press. Merritt AD, Kang KW, Conneally PM, Gersting JM, Rigo T (1976): MEGADATS: A computer system for family data acquisition, storage, and analysis. In Emery AEH, Miller JR (eds): “Registers for the Prevention and Detection of Genetic Disease.” New York: Symposia Specialists, pp 31 -51. Miller JR (1971): Statement regarding the confidentiality of records in surveillance and registry programs. In Hook EB, Janerich DT, Porter IH (eds): “Monitoring, Birth Defects and Environment - The Problem of Surveillance..’ New York: Academic Press. pp 301 -302. Miller JR (1975): Noteson the B.C. registry for handicapped children and adults. J Clin Comp 4:234-243. Mitchell JA, Loughman WD,Epstein CJ (1980): GENFILES: A computerized medical genetics information network. 11. MEDGEN: The clinical genetics system. Am J Med Genet (this volume). Mitchell JA, Epstein CJ, Loughman WD, Carey JC, Sherman S (1979): A computerized information system for clinical genetics. Am J Hum Genet 31:79A. Moores HM (1976): Computing problems in genetic registers, In Emery AEH, Miller JR (eds): “Registers for the Prevention and Detcction of Genetic Disease.” New York: Symposia Specialists, pp 115-121. Murphy EA (1975): Organization of genetic data. J Clin Comp 4:244-247. National Research Council (Committee on lnorn Errors of Metabolism) (1975): Registries of genetic disease and disability, and family screening. In “Genetic Screening.” Washington, DC: National Academy of Sciences, pp 150--157. Parisconference (1971): Standardization in Human Cytogcnctics. Birth Defects Orig Art Ser 8(7). Paris Conference (1971): Supplement (1 975): Standardization in Human Cytogenetics, Birth Defects Orig Art Ser 9(9). Prescott GH, Rivas ML, Shanbeck L, Macfarlane DW, Wyandt HE, Breg WR, Lubs HA, Magenis RE, Summit RL, Palmer CG, Hecht F, Kimberling W, Clow D (1978): The Interregional Cytogenetic Register System (ICRS). New York: Birth Defects: Orig Art Ser 14(6C):269-279. The National Foundation-March of Dimes. 278 Loughman, Mosher, and Epstein Pruzansky S, Parris P, Laffer J (1976): The making o f a data bank for the Center for Craniofacial Anomalies. In McCormick BH (ed): “Proceedings of the Third Illinois Conference on Medical Information Systems.” Chicago: University of Chicago Press, pp 267-277. Rimoin DL, Crow JF, Miller JR, Porter I, Rosenberg L, Rivas M, Saiver S, Motulsky A (1977): Report of thc Ad Hoc Committee of the American Society of Human Genetics on the Accreditation of Medical Genetics Services. Program and Abstracts of the 28th Annual Meeting. The American Society of Human Genetics, October 21, 1977. Science News (1 76): The National Foundation-March of Dimes, Science Information Division, news release, December 4, 1976. Starkweather, JA (1 978): Personal communication. Trimble BK (1976): Record linkage and genetic registers. In Emery AEN, Miller JR (eds): “Registers for the Prevention and Detection of Genetic Disease.” New York: Symposia Specialists, pp 87-101. Vlietinck RF, Van den Berghe H (1976): The Belgian National Genetic Register. In Emery AEH, Miner JR (cds): “Registers for thc Prevention and Detection of Genetic Disease.” New York: Symposia Specialists, pp 65-71.