Corrigendum 

Association and linkage analyses of RGS4 polymorphisms in schizophrenia 

Kodavali V Chowdari, Karoly Mirnics, Prachi Semwal, Joel Wood, Elizabeth Lawrence, Triptish Bhatia, Smita N Deshpande, Thelma BK, Robert E Ferrell, Frank A Middleton, Bernie Devlin, Pat Levitt, David A Lewis and Vishwajit L Nimgaonkar 

Human Molecular Genetics 11, 1373-1380 (2002) 

The original manuscript reported on association and linkage between polymorphisms of the RGS4 gene and schizophrenia. The primary analyses suggested association and linkage with schizophrenia.  Secondary analyses using a sample of patients with bipolar I disorder (BD1) suggested significant overall differences in estimated RGS4 haplotype frequency between the BD1 cases and two groups of population based controls. 

We recently noted a data entry error restricted to genotypes for SNP7 among the BD1 cases and their parents. The corrections presented below have no bearing on the original reported associations with schizophrenia.  They clarify a puzzling variation noted earlier between the BD1 and schizophrenia samples. 

Unlike the prior results, the revised genotypes now show no significant haplotype frequency differences between the BD1 cases and either group of controls.  The initial and the revised haplotype distributions are provided (Table 2R, below).  Unchanged values for the population based controls, listed in Table 2 in the original paper are also presented for comparison.  Due to the revisions, the estimates for linkage disequilibrium with other RGS4 SNPs are now larger (see revised data below).  Our estimates for transmission distortion in the BD1 sample using TRANSMIT software are lower (c² = 12.94, 10 df, p = 0.2269).  However, the overall probability for biased parental transmission of haplotypes bearing SNPs 1, 4, 7 and 18 among the family-based schizophrenia and the bipolar samples continues to be significant (p = 0.0033).

The authors regret this error.

Table 2R.  Estimated haplotype frequencies among cases and unrelated controls 

Haplotype	Pittsburgh BD1 cases (old)	Pittsburgh BD1 cases (revised)	Neonatal controls	Adult controls
SNP 1ľ4ľ7ľ18
oľoľoľo	0.044	0.067	0.096	0.066
·ľoľoľo	0.029	0.017	0.004	0.021
oľ·ľoľo	0	0	0.006	0.006
·ľ·ľoľo	0	0	0	0
oľoľ·ľo	0.025	0	0	0
·ľoľ·ľo	0.378	0.387	0.388	0.442
oľ·ľ·ľo	0	0	0	0.006
·ľ·ľ·ľo	0	0	0	0
oľoľoľ·	0.017	0.015	0	0.004
·ľoľoľ·	0	0	0	0.006
oľ·ľoľ·	0.349	0.459	0.439	0.425
·ľ·ľoľ·	0.007	0.006	0.008	0.013
oľoľ·ľ·	0	0	0	0
·ľoľ·ľ·	0.050	0.049	0.053	0.013
oľ·ľ·ľ·	0.102	0	0.006	0.000
·ľ·ľ·ľ·	0	0	0	0

The Caucasian cases from Pittsburgh (schizophrenia, n = 93, and bipolar I disorder, BD1, n = 82) were compared separately with unscreened Caucasian neonatal or adult controls from Pittsburgh (n = 85 and 89, respectively).  An omnibus test based on likelihood ratios detected overall differences in haplotype frequencies between the BD1 cases and each group of controls (p = 0.0002) in the original publication.  Significant case-control differences are not detectable using the revised data. 

For each SNP, 'o' represents allele 1 and '·' represents allele 2.

Table 3R.  Pair-wise LD among cases, parents and unrelated controls 

SNP pairs	Pittsburgh Bipolar I disorder (old)		Pittsburgh Bipolar I disorder (revised)
	cases	parents	cases	parents
1&4	0.971	0.951	0.971	0.951
1&7	0.846	0.863	1.0	1.0
1&18	0.753	0.780	0.753	0.780
4&7	0.620	0.537	1.0	0.997
4&18	1	0.959	1.0	0.959
7&18	0.646	0.611	0.765	0.808
average	0.806	0.784	0.853	0.915

LD was estimated using D’ (19) .  The values reported in our manuscript are presented along with the revised estimates.  Values that changed are bolded.  Significant LD was observed for all pair-wise calculations (D’ > 0.78; p < 0.00013).