1: \begin{abstract}

2: Genetical genomics experiments have now been routinely conducted to

3: measure both the genetic markers and gene expression data on the same

4: subjects. The gene expression levels are often treated as quantitative

5: traits and are subject to standard genetic analysis in order to

6: identify the gene expression quantitative loci (eQTL). However, the

7: genetic architecture for many gene expressions may be complex, and

8: poorly estimated genetic architecture may compromise the inferences of

9: the dependency structures of the genes at the transcriptional level. In

10: this paper we introduce a sparse conditional Gaussian graphical model

11: for studying the conditional independent relationships among a set of

12: gene expressions adjusting for possible genetic effects where the gene

13: expressions are modeled with seemingly unrelated regressions. We

14: present an efficient coordinate descent algorithm to obtain the

15: penalized estimation of both the regression coefficients and the

16: sparse concentration matrix. The corresponding graph can be used to

17: determine the conditional independence among a group of genes while

18: adjusting for shared genetic effects. Simulation experiments and

19: asymptotic convergence rates and sparsistency are used to justify our

20: proposed methods. By sparsistency, we mean the property that all

21: parameters that are zero are actually estimated as zero with

22: probability tending to one. We apply our methods to the analysis of a

23: yeast eQTL data set and demonstrate that the conditional Gaussian

24: graphical model leads to a more interpretable gene network than a

25: standard Gaussian graphical model based on gene expression data alone.

26: \end{abstract}