HLA population genetics: a Lebanese population

Human leukocyte antigen (HLA) typing was done in 426 Lebanese subjects of 88 families, in which 347 haplotypes were identified. The A, B, C, DRB1, DRB3/4/5, DQB1 and DPB1 loci were typed at high resolution. This study shows that information theory, as originally developed by Claude Shannon in 1948,...

Full description

Saved in:
Bibliographic Details
Published inTissue antigens Vol. 80; no. 4; pp. 341 - 355
Main Authors Cano, Pedro, Testi, Manuela, Andreani, Marco, Khoriaty, Evelyne, Monsef, Jad Bou, Galluccio, Tiziana, Troiano, Maria, Fernandez-Vina, Marcelo, Inati, Adlette
Format Journal Article
LanguageEnglish
Published Oxford, UK Blackwell Publishing Ltd 01.10.2012
Subjects
Alleles
Drb1 protein
Ethnic groups
Ethnic Groups - genetics
Female
Gene Frequency
Genetic Variation - immunology
Genetics, Population
Haplotypes
Histocompatibility antigen HLA
Histocompatibility Testing
HLA Antigens - classification
HLA Antigens - genetics
HLA Antigens - immunology
human leukocyte antigen
Humans
Information Theory
Lebanese
Lebanon
Linkage Disequilibrium
Male
Molecular Typing
Population genetics
Tissue typing
Lebanon
Online AccessGet full text

Cover

More Information
Summary:Human leukocyte antigen (HLA) typing was done in 426 Lebanese subjects of 88 families, in which 347 haplotypes were identified. The A, B, C, DRB1, DRB3/4/5, DQB1 and DPB1 loci were typed at high resolution. This study shows that information theory, as originally developed by Claude Shannon in 1948, provides a promising theoretical foundation to study the population genetics of a genetic system like HLA. Although Lebanese carry HLA alleles found in other populations, the association of these alleles into haplotypes is quite unique. Comparisons are made with the main ethnic groups. Two haplotypes well represented in the Lebanese population are not identified in any global population: L1 = {A*26:01:01 ‐ B*35:01:01:01‐ C*04:01:01:01‐ DRB1*16:01:01 ‐ DRB5*02:02 ‐ DQB1*05:02:01} and L2 = {A*02:02 ‐ B*41:01‐ C*17:01:01:01 ‐DRB1*11:04:01 ‐ DRB3*02:02:01:01‐ DQB1*03:01:01:01}. By studying linkage disequilibrium in two blocks at a time, with the division of the blocks at different levels in consecutive cycles, conserved haplotypes in full linkage disequilibrium come to light, such as {A*26:01:01‐ B*35:01:01:01 ‐ C*04:01:01:01 ‐ DRB1*16:01:01 ‐ DRB5*02:02 ‐ DQB1*05:02:01‐ DPB1*03:01:01} and {A*33:01:01 ‐ B*14:02:01 ‐ C*08:02:01 ‐ DRB1*01:02:01‐ DQB1*05:01:01:01 ‐ DPB1*04:01:01:01}.
Bibliography:Appendix S1 Haplotypes and complete linkage disequilibrium data.Table S1 HLA-A allele frequencies in a Lebanese populationTable S2 HLA-B allele frequencies in a Lebanese populationTable S3 HLA-C allele frequencies in a Lebanese populationTable S4 HLA-DRB1 allele frequencies in a Lebanese populationTable S5 HLA-DRB3/4/5 allele frequencies in a Lebanese populationTable S6 HLA-DQB1 allele frequencies in a Lebanese populationTable S7 HLA-DPB1 allele frequencies in a Lebanese populationTable S8 HLA genetic diversity in Lebanese compared to European (comparison of the homozygosity ration to entropy as measures of genetic diversity.)Table S9 Linkage disequilibrium in a Lebanese population: B vs CTable S10 Linkage disequilibrium in a Lebanese population: A vs B-CTable S11 Linkage disequilibrium in a Lebanese population: DRB1 vs DQB1Table S12 Linkage disequilibrium in a Lebanese population: B-C vs DRB1-DQB1 (including DRB1-DRB3/4/5-DQB1)Table S13 Linkage disequilibrium in a Lebanese population: A-B-C vs DRB1-DQB1Table S14 Linkage disequilibrium in a Lebanese population: A vs B-C-DRB1-DQB1Table S15 Linkage disequilibrium in a Lebanese population: DRB1-DQB1 vs DPB1Table S16 Linkage disequilibrium in a Lebanese population: A-B-C vs DRB1-DQB1-DPB1Table S17 Linkage disequilibrium in a Lebanese population: A vs B-C-DRB1-DQB1-DPB1Table S18 Linkage disequilibrium in a Lebanese population: A-B-C-DRB1-DQB1 vs DPB1Table S19 Significant allele frequency differences between Lebanese and European populationsTable S20 Significant B-C block frequency differences between Lebanese and European populationsTable S21 Significant DR-DQ block frequency differences between Lebanese and European populationsTable S22 Comparison of haplotype frequencies in Lebanese and other ethnic groups. Frequencies are presented as the negative logarithm of the frequency. High frequency in green. Unique Lebanese haplotypes highlighted in brownTable S23 Typing ambiguities not resolvedFigure S1 Comparison of homozygosity and entropy for individual alleles. There is good correlation between the two measures, except when the homozygosity is less than 20%.Figure S2 Entropy as a function of allele frequency (compare with Figure S3)Figure S3 Homozygosity as a function of allele frequency (compare with Figure S2)Figure S4 Entropy growth with the number of alleles. The blue line shows the line of maximum entropy, which occurs when the alleles are evenly distributed. The orange line shows a typical HLA locus, HLA-B in this case. As the number of alleles identified grows, their contribution of total entropy and to genetic diversity quickly slows down becoming irrelevant (here the entropy is calculated with logs in base 2 and a constant of 15).Figure S5 Circular relationship of delta prime (D′) and mutual information I(X;Y) when cell a (X ∧ Y) in 2 × 2 table increases progressively with constant b (X ∧ ¬Y) , c (¬X ∧ Y) and d (¬X ∧ ¬Y) . Both D′ and I(X;Y) first increase and then decrease as the values in cell a progressively increase, but they do it at different rates and in different fashion. This up and down behaviour is caused by fact that the effect of the increase in a depends on the increase of d, if d is constant, there is a point at which the effect of the increase of a disappears. In order to assess the usefulness of D′ and I(X;Y) as measures of linkage disequilibrium one can compare point P to point Q, and then point R to point S, and their corresponding 2 × 2 tables. Is it more important to measure the increase in linkage disequilibrium when moving from point L to point M, or is rather more important to measure the increase when moving from point L to point Q? Which is the point of maximum linkage disequilibrium, point M or point Q?Figure S6 In order to assist in the answer to the questions presented in Figure S7, we include here two other measures of association that could also be used to measure linkage disequilibrium, the statistic chi square and the Ochiai geometric distance measure. Although point Q seems to have more weight than point P in terms of association and therefore linkage disequilibrium, and therefore favouring mutual information over delta prime as a measure of linkage disequilibrium, it must be emphasise, however, that each measure is actually measuring something different. The main advantage we see with mutual information is that it can be integrated with other information measures into a coherent account of descriptive population genetics given by the equations presented in section 2.4. Shannon's information theory and population genetics.Figure S7 Relationship of chi square and mutual information I(X;Y). In our study, we chose a level of I(X;Y) > 0.1 to screen for associations between alleles or blocks of alleles that could show relevant linkage disequilibrium.
ArticleID:TAN1936
ark:/67375/WNG-D86X9J27-9
istex:B08276A993D5EF19C6497524BAA3615C37697D49
ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ObjectType-Article-2
ObjectType-Feature-1
ISSN:0001-2815
1399-0039
1399-0039
DOI:10.1111/j.1399-0039.2012.01936.x