Genome scan meta-analysis of schizophrenia and bipolar disorder, part I: Methods and power analysis - PubMed (original) (raw)

Genome scan meta-analysis of schizophrenia and bipolar disorder, part I: Methods and power analysis

Douglas F Levinson et al. Am J Hum Genet. 2003 Jul.

Abstract

This is the first of three articles on a meta-analysis of genome scans of schizophrenia (SCZ) and bipolar disorder (BPD) that uses the rank-based genome scan meta-analysis (GSMA) method. Here we used simulation to determine the power of GSMA to detect linkage and to identify thresholds of significance. We simulated replicates resembling the SCZ data set (20 scans; 1,208 pedigrees) and two BPD data sets using very narrow (9 scans; 347 pedigrees) and narrow (14 scans; 512 pedigrees) diagnoses. Samples were approximated by sets of affected sibling pairs with incomplete parental data. Genotypes were simulated and nonparametric linkage (NPL) scores computed for 20 180-cM chromosomes, each containing six 30-cM bins, with three markers/bin (or two, for some scans). Genomes contained 0, 1, 5, or 10 linked loci, and we assumed relative risk to siblings (lambda(sibs)) values of 1.15, 1.2, 1.3, or 1.4. For each replicate, bins were ranked within-study by maximum NPL scores, and the ranks were averaged (R(avg)) across scans. Analyses were repeated with weighted ranks ((sqrt)N[genotyped cases] for each scan). Two P values were determined for each R(avg): P(AvgRnk) (the pointwise probability) and P(ord) (the probability, given the bin's place in the order of average ranks). GSMA detected linkage with power comparable to or greater than the underlying NPL scores. Weighting for sample size increased power. When no genomewide significant P values were observed, the presence of linkage could be inferred from the number of bins with nominally significant P(AvgRnk), P(ord), or (most powerfully) both. The results suggest that GSMA can detect linkage across multiple genome scans.

PubMed Disclaimer

Figures

Figure 1

Figure 1

Determining _P_ord: observed and expected ordered _R_avg values. The black line shows the 120 observed _R_avg values, sorted with the first-place bin on the right. The light gray line and vertical error bars show the mean ± 2 SD of the _j_th-place bin from 5,000 random permutations. In the absence of linkage, _R_avg values lower than the expected distribution are observed infrequently throughout the genome. With linkage, they are clustered among the highest values of _R_avg, as shown here. As shown in table 1, _P_ord is the probability of observing a given value of _R_avg (black line) by chance, given its place in the order of observed _R_avg values, which is determined from the distribution of ordered _R_avg values (light gray line and error bars).

Figure 2

Figure 2

Marker maps for simulation studies. Genotypes were simulated for 180-cM chromosomes, on each of which six 30-cM bins (segments) were structured as shown. Marker locations are M1, M2, M3. For the SCZ data sets, some studies had 10-cM marker spacing (A) and others had 15-cM spacing (B); all BP data sets were simulated with 10 cM spacing. Linked (disease) and unlinked (nondisease) chromosomes were created (C). Disease loci (indicated by “D”) were located midway between two markers, either in the first (edge) or third (middle) bin. Genomes consisted of 20 chromosomes (120 bins); see table 5 for structures of genomes.

Figure 3

Figure 3

Mean average rank by bin for unlinked replicates. Shown are the means ± SD of the average ranks of each bin (over all chromosomes) in the 1,000 unlinked replicates of the SCZ data set. “Edge” bins (1 and 6) have lower summed ranks than “mid” bins (2–5). This contributed to an inflated type I error rate when _P_ord values were computed by permuting the order of ranks in each study by bin, and was corrected by permuting by chromosome so that summed ranks of edge and middle bins were compared with their own distributions.

Figure 4

Figure 4

Mean number of bins per replicate achieving genomewide significance. Shown is the mean number of bins (in 100 replicates per model) with

_P_⩽.0004167

(the threshold for genomewide significance = .05/120), for weighted analyses. Diamonds represent SCZ data sets, squares represent BP-N data sets, and circles represent BP-VN data sets. Blackened symbols represent data for

λ_sibs_=1.3

, unblackened symbols represent data for

λ_sibs_=1.15

.

Figure 5

Figure 5

Power to detect linked bins (unweighted analyses). Shown are the proportions of bins containing linked loci that achieved

P _AvgRnk_⩽.05

(blackened symbols) or ⩽.01 (unblackened symbols), for each model (see table 5), for simulated replicates of SCZ (A), BP-VN (B) and BP-N (C) data sets. Squares represent md1 replicates, diamonds represent ed1 replicates, circles represent ed1md4 replicates, and triangles represent ed2md8 replicates. See tables 2, 3, and 4 for characteristics of each data set.

_N_=100

per model.

Figure 6

Figure 6

Number of loci with nominally significant _P_AvgRnk, _P_ord, or both. Data from table 7 are shown for the BP-VN (9 studies) and SCZ (20 studies) data sets weighted analysis, with _P_ord computed by permuting by chromosome, for

λ_sibs_=1.15

. Blackened bars represent mean number of bins with both values <.05, diagonally striped bars represent bins with only

P AvgRnk<.05

, and white bars represent bins with only

P ord<.05

. Sets of bars are labeled by data set (BP or SCZ) and the number of linked loci in the data set (1 = md1, 5 = ed1md4, and 10 = ed2md8). Shown are the number of bins with these values for disease + adjacent bins (left-most six sets of bars), bins on chromosomes containing no disease locus (next six sets), and bins in completely unlinked data sets (the average of BP-VN-unlinked and SCZ-unlinked, which were virtually identical). See text for details.

References

Electronic-Database Information

    1. Online Mendelian Inheritance in Man (OMIM), http://www.ncbi.nlm.nih.gov/Omim/ (for SCZD, MAFD1, and MAFD2)

References

    1. ——— (2002a) Meta-analysis of whole-genome linkage scans of bipolar disorder and schizophrenia. Mol Psychiatry 7:405–411 - PubMed
    1. Badner JA, Gershon ES (2002b) Regional meta-analysis of published data supports linkage of autism with markers on chromosome 7. Mol Psychiatry 7:56–66 - PubMed
    1. Berrettini WH (2000) Are schizophrenic and bipolar disorders related? A review of family and molecular studies. Biol Psychiatry 48:531–538 - PubMed
    1. Dorr D, Rice J, Armstrong C, Reich T, Blehar M (1997) A meta-analysis of chromosome 18 linkage data for bipolar illness. Genet Epidemiol 14:617–622 - PubMed
    1. Gill M, Vallada H, Collier D, Sham P, Holmans P, Murray R, McGuffin P, et al (1996) A combined analysis of D22S278 marker alleles in affected sib-pairs: support for a susceptibility locus for schizophrenia at chromosome 22q12. Schizophrenia Collaborative Linkage Group (Chromosome 22). Am J Med Genet 67:40–45 - PubMed

Publication types

MeSH terms

LinkOut - more resources