Computational Bioinformatics & Bio-imaging Laboratory (CBIL)


Internal Use

Improved Diagnostics of the Muscular Dystrophies (NS29525-13A)

Aim 1. Development of molecular diagnostics using the existing 107 biopsy data set.

Our general approach to data analyses under the previous award period was to begin small (e.g. pilot studies in single or a few diagnostic groups), learn the source and extent of confounding variables, test sensitivity and specificity of specific analyses using focused RT-PCR and protein pathway studies, and gradually build computational experience both within the laboratory and via key collaborating laboratories. This systematic approach has led to a large series of publications on investigations in muscle and muscle disease (Chen et al. 2000; Tezak et al. 2002; Bakay et al. 2002; Tseng et al. 2002; Fischer et al. 2002; Chen et al. 2002; Borup et al. 2002; Hittel et al. 2003; Steenman et al. 2003; Bey et al. 2003; Seo et al. 2003; Chen et al. 2003; DiGiovanni et al. 2004). This experience in muscle as a model system resulted in the award of key NIH grants to the laboratory in other tissue systems, including spinal cord injury (DiGiovanni et al. 2003), and a NHLBI Programs in Genomic Applications grant that allowed us to develop state-of-the-art public data dissemination tools (Chen et al. 2004; Almon et al. 2003) (see Finally, a key value added project has been an extensive study of muscle regeneration funded by the MDA, including a 27 time point in vivo temporal series (Zhao et al. 2002, Zhao et al. 2003; Zhao et al. 2004; Fulco et al. 2003). The proposed first aim is a direct extension of our preliminary data and ongoing collaborations, where we use our pre-existing 107 muscle biopsy data set as a platform for developing sensitive and specific methods for gene selection and subsequent molecular diagnostics.

Aim 1A. Refine the gene selection using the existing 107 biopsy data set for optimized diagnosis of each dystrophy group of defined etiology.

Aim 1B. Develop diagnostic classifications for each of the groups, tested by profiling 30 biopsies of known diagnosis, and determine classifiers? generalizable performance.

Aim 1C. Build a visual interface with TreeMap as a support of diagnosis, including web implementation of the interface.

Aim 1D. Build an RT-PCR-based mRNA diagnostics method for screening of 200 muscle biopsies of unknown diagnosis. Assign these 200 patients to specific diagnostic categories. The corresponding genes will be sequenced to determine accuracy of the mRNA-based tests.

Aim 2. Gene selection for biochemical pathway building.

As shown in Preliminary Data, we have successfully constructed biochemical pathway models for Emery Dreifuss muscular dystrophy using related nuclear envelope defects (emerin, Lamin A/C), and gene selection methods (leave-one-out based on weighted Fisher criteria, with p values and fold changes relative to normal muscle). In this aim, we will extend this analysis to all 10 diagnostic groups, with our additional knowledge that some proteins directly or indirectly interact (much as the case with Lamin A/C and emerin).

Aim 2A. Expression profile 30 DMD, 10 alpha-sarcoglycan, 10 beta-sarcoglycan, and 30 dysferlin, 30 calpain III, and 30 lamin A/C patients identified as mutation-positive in Aim 1 (140 subjects total). The goal is to increase power of analysis (~10-50 per group).

Aim 2B. We will test the hypothesis that the first five groups share downstream biochemical pathways that will be distinct from normal, and other disease controls (EDMD, FSHD, JDM, AQM; spastin; see Progress Report). We will use gene selection of individual groups and larger order pathophysiological groups (e.g. membrane, nuclear envelope) for nucleation of pathway building.

Aim 2C. Build interacting pathways from a hierarchy of biochemical defects: alpha-sarcoglycan/beta-sarcoglycan-calpain III-dystrophin-dysferlin. The goal here would be to connect the pathways specific to each of the four disorders into a network of pathways.

A future aim will be biological validation, by production of antibodies will be produced against proposed pathway members, affinity columns produced, and column affinity purification methods used to identify interacting proteins and modification states in both normal and dystrophic muscle.

Aim 3. Identification of novel classifications of muscular dystrophy.

As cited above, approximately 160 of the 200 patient muscle biopsies we receive each year are unable to be assigned a diagnosis. This Aim is to identify novel causes of muscular dystrophy using comparisons of individual patients to the large ?data warehouse? obtained through Aim 2.

Aim 3A. 30 sib pairs of patients (n=60 patient biopsies) that have had all known types of muscular dystrophy ruled out by protein and gene sequence analysis will be expression profiled, and tested against the 250 patient profile database of known types. Two profiles each (n=4 for two sibs) will be done from different regions of the muscle biopsy for each subject to control for tissue heterogeneity.

Aim 3B. Class discovery will be done to determine if any subset of the 60 subjects form a cluster among themselves, suggesting a new diagnostic group. Gene selection will be done in this group to identify potential biochemical pathways that may represent the primary biochemical defect. Sequence analyses will be done in those pathway members showing significantly decreased expression, assuming a loss-of-function model (isolated cases).

Aim 3C. We will test the hypothesis that significantly down-regulated transcripts (3 standard deviations from mean of remainder of database) reflect loss-of-function mutations in the corresponding gene. This will be done for both individual subjects, and any cluster of subjects that may be found in Aim 3B. Sequence analysis of the candidate genes will be done to verify the novel diagnostic categories.






Copyright ©2004, Computational Bioinformatics and Bioimaging Laboratory (CBIL), Advanced Research Institute, Virginia Tech.

Last Updated: 03/03/2009. Suggestions/Comments - Webmaster