Population dynamics of potentially harmful haplotypes: a pedigree analysis.

BACKGROUND: The identification of low-frequency haplotypes, never observed in homozygous state in a population, is considered informative on the presence of potentially harmful alleles (candidate alleles), putatively involved in inbreeding depression. Although identification of candidate alleles is challenging, studies analyzing the dynamics of potentially harmful alleles are lacking. A pedigree of the highly endangered Gochu Asturcelta pig breed, including 471 individuals belonging to 51 different families with at least 5 offspring each, was genotyped using the Axiom PigHDv1 Array (658,692 SNPs). Analyses were carried out on four different cohorts defined according to pedigree depth and at the whole population (WP) level. RESULTS: The 4,470 Linkage Blocks (LB) identified in the Base Population (10 individuals), gathered a total of 16,981 alleles in the WP. Up to 5,466 (32%) haplotypes were statistically considered candidate alleles, 3,995 of them (73%) having one copy only. The number of alleles and candidate alleles varied across cohorts according to sample size. Up to 4,610 of the alleles identified in the WP (27% of the total) were present in one cohort only. Parentage analysis identified a total of 67,742 parent-offspring incompatibilities. The number of mismatches varied according to family size. Parent-offspring inconsistencies were identified in 98.2% of the candidate alleles and 100% of the LB in which they were located. Segregation analyses informed that most potential candidate alleles appeared de novo in the pedigree. Only 17 candidate alleles were identified in the boar, sow, and paternal and maternal grandparents and were considered segregants. CONCLUSIONS: Our results suggest that neither mutation nor recombination are the major forces causing the apparition of candidate alleles. Their occurrence is more likely caused by Allele-Drop-In events due to SNP calling errors. New alleles appear when wrongly called SNPs are used to construct haplotypes. The presence of candidate alleles in either parents or grandparents of the carrier individuals does not ensure that they are true alleles. Minimum Allele Frequency thresholds may remove informative alleles. Only fully segregant candidate alleles should be considered potentially harmful alleles. A set of 16 candidate genes, potentially involved in inbreeding depression, is described.

Approaching autozygosity in a small pedigree of Gochu Asturcelta pigs.

BACKGROUND: In spite of the availability of single nucleotide polymorphism (SNP) array data, differentiation between observed homozygosity and that caused by mating between relatives (autozygosity) introduces major difficulties. Homozygosity estimators show large variation due to different causes, namely, Mendelian sampling, population structure, and differences among chromosomes. Therefore, the ascertainment of how inbreeding is reflected in the genome is still an issue. The aim of this research was to study the usefulness of genomic information for the assessment of genetic diversity in the highly endangered Gochu Asturcelta pig breed. Pedigree depth varied from 0 (founders) to 4 equivalent discrete generations (t). Four homozygosity parameters (runs of homozygosity, FROH; heterozygosity-rich regions, FHRR; Li and Horvitz's, FLH; and Yang and colleague's FYAN) were computed for each individual, adjusted for the variability in the base population (BP; six individuals) and further jackknifed over autosomes. Individual increases in homozygosity (depending on t) and increases in pairwise homozygosity (i.e., increase in the parents' mean) were computed for each individual in the pedigree, and effective population size (Ne) was computed for five subpopulations (cohorts). Genealogical parameters (individual inbreeding, individual increase in inbreeding, and Ne) were used for comparisons. RESULTS: The mean F was 0.120 ± 0.074 and the mean BP-adjusted homozygosity ranged from 0.099 ± 0.081 (FLH) to 0.152 ± 0.075 (FYAN). After jackknifing, the mean values were slightly lower. The increase in pairwise homozygosity tended to be twofold higher than the corresponding individual increase in homozygosity values. When compared with genealogical estimates, estimates of Ne obtained using FYAN tended to have low root-mean-squared errors. However, Ne estimates based on increases in pairwise homozygosity using both FROH and FHRR estimates of genomic inbreeding had lower root-mean-squared errors. CONCLUSIONS: Parameters characterizing homozygosity may not accurately depict losses of variability in small populations in which breeding policy prohibits matings between close relatives. After BP adjustment, the performance of FROH and FHRR was highly consistent. Assuming that an increase in homozygosity depends only on pedigree depth can lead to underestimating it in populations with shallow pedigrees. An increase in pairwise homozygosity computed from either FROH or FHRR is a promising approach for characterizing autozygosity.

Copy Number Variation Regions Differing in Segregation Patterns Span Different Sets of Genes.

Copy number variations regions (CNVRs) can be classified either as segregating, when found in both parents, and offspring, or non-segregating. A total of 65 segregating and 31 non-segregating CNVRs identified in at least 10 individuals within a dense pedigree of the Gochu Asturcelta pig breed was subjected to enrichment and functional annotation analyses to ascertain their functional independence and importance. Enrichment analyses allowed us to annotate 1018 and 351 candidate genes within the bounds of the segregating and non-segregating CNVRs, respectively. The information retrieved suggested that the candidate genes spanned by segregating and non-segregating CNVRs were functionally independent. Functional annotation analyses allowed us to identify nine different significantly enriched functional annotation clusters (ACs) in segregating CNVR candidate genes mainly involved in immunity and regulation of the cell cycle. Up to five significantly enriched ACs, mainly involved in reproduction and meat quality, were identified in non-segregating CNVRs. The current analysis fits with previous reports suggesting that segregating CNVRs would explain performance at the population level, whereas non-segregating CNVRs could explain between-individuals differences in performance.

Segregation patterns and inheritance rate of copy number variations regions assessed in a Gochu Asturcelta pig pedigree.

Copy Number Variation Regions (CNVR) were subjected to pedigree analysis to contribute to the understanding of their segregation patterns. Up to 492 Gochu Asturcelta pig individuals forming 478 different parents-offspring trios (61 different families) were genotyped using the Axiom_PigHDv1 Array (658,692 SNPs). CNVR calling, performed using two different platforms (PennCNV and QuantiSNP), allowed to identify a total of 344 candidate CNVR on the 18 porcine autosomes covering about 106.8 Mb of the pig genome. Sixty-nine CNVR were identified, to some extent, in both the parents and the offspring and were classified as segregating CNVR. The other candidate CNVR were called in one or more progeny but in neither parent and classified either as singleton or recurrent de novo CNVR. Segregating CNVR were, on average, larger and more frequent than the recurrent de novo CNVR (444.8 kb vs 287.9 kb long and 34 vs 5 individuals, respectively). In any case, segregating CNVR did not conform to strict Mendelian inheritance patterns: estimates of average paternal and maternal transmission rates ranged from 11.0 % to 13.4 % and mean inheritance rate was below 21 %. This issue should be carefully considered when interpreting the results of CNV studies. Segregating CNVR, present across generations, are unlikely to be artifacts or false positives and can be hypothesized to be important at the population level.

Understanding Mendelian errors in SNP arrays data using a Gochu Asturcelta pig pedigree: genomic alterations, family size and calling errors.

Up to 478 Gochu Asturcelta pig parents-offspring trios (61 different families) were genotyped using the Axiom_PigHDv1 Array to identify the causes of Mendelian errors (ME). Up to 545,364 SNPs were retained. Up to 40,540 SNPs gathering 292,297 allelic mismatches were identified and were overlapped with SINEs and LINEs (Sscrofa genome 11.1). Copy number variations (CNV) were called using PennCNV. ME were classified into eight different classes according to the trio member ("Trio" meaning no assignment) and the allele on which ME was identified: TrioA/B, FatherA/B, MotherA/B, OffspringA/B. Most ME occurred due to systematic causes: (a) those assigned to the Father, Mother or Offspring occurred by null or partial null alleles characterized by heterozygote deficiency, varied with family size, involved a low number of loci (6506), and gathered most mismatches (228,145); (b) TrioB errors varied with family size, covaried with SINEs, LINEs and CNV, and involved most ME loci (33,483) and mismatches (65,682); and (c) TrioA errors were non-systematic ME with no sampling bias involving 1.2% of mismatches only and a low number of loci (1939). The influence of TrioB errors on the overall genotyping quality may be low and, since CNV vary among populations, their removal should be considered in each particular dataset. ME assignable to the Father, Mother or Offspring may be consistent within technological platforms and may bias severely linkage or association studies. Most ME caused by null or partial null alleles can be removed using heterozygote deficiency without affecting the size of the datasets.

Identification and Characterization of Copy Number Variations Regions in West African Taurine Cattle.

A total of 106 West African taurine cattle belonging to the Lagunaire breed of Benin (33), the N'Dama population of Burkina Faso (48), and N'Dama cattle sampled in Congo (25) were analyzed for Copy Number Variations (CNVs) using the BovineHDBeadChip of Illumina and two different CNV calling programs: PennCNV and QuantiSNP. Furthermore, 89 West African zebu samples (Bororo cattle of Mali and Zebu Peul sampled in Benin and Burkina Faso) were used as an outgroup to ensure that analyses reflect the taurine cattle genomic background. Analyses identified 307 taurine-specific CNV regions (CNVRs), covering about 56 Mb on all bovine autosomes. Gene annotation enrichment analysis identified a total of 840 candidate genes on 168 taurine-specific CNVRs. Three different statistically significant functional term annotation clusters (from ACt1 to ACt3) involved in the immune function were identified: ACt1 includes genes encoding lipocalins, proteins involved in the modulation of immune response and allergy; ACt2 includes genes encoding coding B-box-type zinc finger proteins and butyrophilins, involved in innate immune processes; and Act3 includes genes encoding lectin receptors, involved in the inflammatory responses to pathogens and B- and T-cell differentiation. The overlap between taurine-specific CNVRs and QTL regions associated with trypanotolerant response and tick-resistance was relatively low, suggesting that the mechanisms underlying such traits may not be determined by CNV alterations. However, four taurine-specific CNVRs overlapped with QTL regions associated with both traits on BTA23, therefore suggesting that CNV alterations in major histocompatibility complex (MHC) genes can partially explain the existence of genetic mechanisms shared between trypanotolerance and tick resistance in cattle. This research contributes to the understanding of the genomic features of West African taurine cattle.