Download datasets and supplementary data files |
---|
Summary
Data generated by RW Williams, Lu Lu, Manjunatha Jagalur, and David Kulp. All arrays run at the VA Medical Center, Memphis, by Yan Jiao.
Data entered by Arthur Centeno and Manju Jagalur, Feb 27, 2009. This data set modified data for two BXD strains. Data were added for BXD79 that had been incorrectly included as a striatum sample (this data set was therefore deleted from the Exon 1.0ST striatum data set). We also changed data for BXD39. As expected, this addition and correction improved QTL mapping values. For example, for Kcnj9 probe set 4519178 the LRS values increased from 103.3 in the Aug08 data to 115.7 for these Feb09 data. Rob is concerned about the high error term of BXD39.
About cases
|
About data processing
The following steps were applied to refine the data by M. Jagalur in RWW lab:
- Strain correction: In this step the strong probe level cis-QTLs were identified and using an expectation maximization (EM)-like method, the genotypes of each marker was re-assigned. This set of reassigned markers was compared to existing list genotypes of BXD strains and the maximal match was identified as the correct strain.
- Sex correction: In this step probes that are highly correlated to sex were identified and using an EM-like method we detected and corrected the sex of single array data sets.
- Data exclusion criteria: In this step individual arrays were evaluated. Arrays were systematically excluded from the data set (drop one out) and the number of cis-QTLs was recomputed. If excluding an array resulted in s significantly higher number of cis-QTLs then that array was considered to be of poor quality and was excluded from the final data set This step was repeated across all arrays in multiple cycles until there was no improvement in number of cis-QTLs.
- Tissue correction: In this step probes that are highly correlated to tissue type were identified and EM-like method was used to identify correct tissue.
- Noise Removal: A noise component was calculated using the expression of "unhybridized"probes (those with the lowest signal) and was removed from the data.
Acknowledgment
We thank Affymetrix Inc. for their generous support of this project and array data set.