A. Data Organization and Biological Classification (Figure 1A)

   From the individual measurements show near the bottom of figure 1A (blue box), summary tables (tan box) are generated that compare the group mean for the KOMP line to the running controls (ZZZ line, see performance of our measurement tools).  It is at this step that a decision is made to flag a measurement that will be considered as abnormal.  Although this abnormality-flag is initially made by a computer algorithm, all values are reviewed by the KOMP team to ensure that call is justified. 
   The abnormal measurement flags are then grouped in the analysis table (green box in figure 1A) according to structural and metabolic processes that have biological and functional meaning.  The four major classifications are:

   1. Trabecular architecture of the femur or vertebra - The primary screen of a KOMP line by µCT is the entry point for more detailed studies by histomorphometry.  The trabecular bone volume measurement is dependent on the trabecular thickness or number value.  This value is affected by the total bone volume (a measurement of size), which is also used as a searchable term.  Secondary abnormalities of the trabeculae can include alterations in the plate/rod like structures of the trabeculae (shape) and/or a difference in mineral density of the trabeculae (density).  Our experience to date has not observed alterations of structure or density independent of the BV/TV measurement. 
    KOMP lines that trigger our histomorphometric study results in measurements for BV/TV, trabecular number and thickness from the histological sections.  Although there is an impressive correlation between µCT and histomorphometry (link back to previous graphs), there can be discrepancies in individual KOMP lines.   At present we do not know if there is an explanation why these differences occur.  We will place greater confidence in the µCT measurements.
   The KOMP lines are studied at 3 months of age when femoral trabecular bone has reached its peak BV/TV after which it is lost, especially in female mice.  Variation in BV/TV may reflect a primary genetic impact on trabecular bone mass or the rate at which bone is accrued or lost.  These differences may be more prominent in females than males giving rise to a gender dimorphism.  The BV/TV assessment in the vertebra is more stable with age, which may account smaller difference in this measure between the genders.  Since axial and appendicular bone have a different embryonic origin, we are noting when there are gender dimorphic differences at both sites, or if the differences are similar in both sexes but different in the femur and vertebra (axial/appendicular dimorphism).  Experience with increasing number of KOMP lines may provide more insight into these dimorphic differences.

    2. Cortical bone architecture of the femur – The cortical bone of the femur carries the majority of the load of the femur and maintains its ability to remodel in response to an underlying genetic defect in bone quality.  The primary measures are the thickness and the size (perimeter) of the cortical bone, which reflect the inherent osteogenic/osteoclastic cells to make cortical bone.  The matrix composition is reflected by the density and porosity measurement, while remodeling activity can be observed in the elliptical shape of the diaphysis (min, max).   The later measures may be abnormal in genetic disorders that affect bone quality.  
    Histomorphometry provides further insight into the cortical bone from the measurement of osteocyte density.  There is a clear sexual dimorphic difference (females greater), but we have not yet observed a genetic difference.  Measurements of bone formation, resorption and remodeling of the endosteal or periosteal cortical surfaces will be implemented in the near future.   

    3. Cellular activity of the trabecular bone of the femur and vertebra – The histomorphometric studies are designed to provide an indication of the type of alteration in the formation/resorption/ remodeling activity that could account for the architectural observations.   The ability to associate features of formation and resorption on the same section provides measures that cannot be made or verified in traditional histology.  Application of these measures across numerous control and KOMP lines will ultimately prove their utility in assessing the cellular activity affecting bone architecture.  The major components of the analysis are:
    a. Bone forming activity – This category identifies surfaces where formation, not associated with remodeling is occurring.  It is composed of 2 measurements, which usually change in parallel.
    Mineral apposition – The temporal incorporation of the mineralization dyes provides insight into the surfaces that are actively depositing new mineralized matrix.  Either the percentage of the trabecular surface that is active (MS/BS) or the thickness of matrix that is produced (MAR) or both can be affected by genetic factors. 
    Osteoblastic activity – The total AP/BS measurement is partitioned to either the AP overlying a labeling surface per bone surface (AP_L/BS) or the proportion of total AP that overlie a labeling surface (AP_L/AP).  Lining cells, which are considered to be quiescent, are identified as AP_NL/BS or AP_NL/AP.

    b. Bone resorbing activity – This category identifies surfaces where resorption, not associated with formation, is occurring.  The total TRAP/BS measurement is partitioned as TRAP activity overlying a non-labeling surface per bone surface (TRAP_NL/BS)or the proportion of total TRAP activity that overlies a non-labeling surface (TRAP_NL/TRAP). 

    4. Whole animal morphometry – (not yet implemented within the portal).  Within the variation in body size of the running controls, we have not observed a correlation between body weight and the BV/TV measurement.  However we expect that systemic disease that impacts somatic growth will also alter bone morphometry including femur length.  The underlying disease may affect fat and muscle mass as well bone mineralization, which is reflected in the BMD measurement.  These values have been imported from in the primary screen of the KOMP pipeline into our tables.  Likewise chondrodystrophies affecting the growth plate may affect limb length with less of an affect on body weight.  Animal lines that exhibit significant differences in whole body measurements will be identified and segregated from other KOMP lines that have normal whole body morphometry. 

    5. Biological Classification
    The table of searchable features for each analyzed KOMP line is presented on the selection table (yellow box) in figure 1A.  The table indication if any skeletal measurement was found (Y or N).  Based on the totality of the accumulated skeletal measures as classified in section A #1, 2 and 3)  the KOMP staff will assign a biological mechanism to explain the skeletal phenotype of an individual KOMP line.  A coding schema has been developed for facilitating computer searches:
Primary search term is base on the Bone Architecture Classification as obtained from µCT and static histomorphometric measure.  It is based on the trabecular bone volume and total volumetric size of the bone (femur and/or vertebra), each of which can have 3 states (low, normal or high).   The final phenotyping code captures both measurements for a total of 9 possibilities (see table 1). 

   The secondary search term is based on the three Cellular Phenotyping Classification: Formation, Resorption and Remodeling.  Each classification has three states (low, normal or high) for a total of 27 possible phenotypes (see table 2).  They are abbreviated as formation (LF, NF, HF), resorption or TRAP activity (LT,NT,HT) and remodeling (LR,NR,HR).

   The composite phenotype code joins the architectural and cellular states that are separated by a colon.  It will allow a search for either phenotypic state separately or in combination (see table 3).

B. Navigating the Web Site (Figure 1B)

   To view data from the bonebase home page, select to see KOMP data, which takes the viewer to a webbed version of our database.  Because this section of the webportal is dynamic, it may take a few moments to load.  If you experience difficulties with loading the file, click the troubleshooting link for suggestions.  Frequently there are cache issues that need to be cleared for the file to work.
    The web-version of the database presents the data in the reverse order of the organizational design.  The site begins with the selection page (figure 5, equivalent to the yellow box in figure 1B) in which the gene symbol (1) and formal gene name (2) are listed.  The table allows the user to search the database for all the lines currently under analysis or completed (3) and whether an abnormality (Y/N) has been identified (4).  The search can be constrained to lines that have been assigned to a specific biological classification (5).  A search feature to specify a specific classification code will be introduce in the near future.  The search can also extract KO lines that show either a gender (6) and/or axial/appendicular dimorphism (7).  The page shows the best phenotype ontological term that resides in the mammalian phenotype (MPI) ontology web portal (8), which allows to user to search our site using these standardized terms.  (http://www.informatics.jax.org/searches/MP_form.shtml).  Links of the gene to the MGI gene and mouse portal and the recently developed international mouse phenotyping consortium (IMPG, the primary portal for KOMP) are also provided (10).  The latter site provides the result of the global phenotyping that has been identified to date on that KO line.  It includes the total body BMD and BMC measurements obtained by Dexa.

   Once a line of interest is identified, the user can drill down into the details of the analysis.  Click the Analysis Result tab (9) on the corresponding KO line and the Analysis Page (figure 6, equivalent to the green box in figure 1B) is activated.  Each category (#1, architecture and cellular activity for femur and vertebra) that contains an abnormal skeletal measurement is listed.  For example, the cellular activity category has abnormal values for all three components (formation, resorption and remodeling).  The table indicates whether the measurement was either high or low (#2) , and present in both genders (#3 as B, M, F, M>F, F>M) and in the axial/appendicular skeleton (#4 as B, Ax, Ap, Ap>Ax, Ax>Ap).     Further to the right of each abnormal value are fields that are currently captured by the MP ontology as (#5) and thus searchable.  The other fields are for future use for ontologies that are more representative of bone biology.  Eventually the website should be searchable for specific µCT or histomorphological measurements.
You can navigate to the previous selection page by clicking #8 or drill down further into the detailed results by clicking button #6, view category summary data.

   Clicking on View Category Summary Data at the far right of each category (#6) opens the summary tables (figures 7A and B, equivalent to the tan box in figure 1B). You can also choose to see the summary data for all the categories by clicking the View Summary Data for All Categories tab located on the top far right. In either case, the mean value of the measurements associated with the KOMP line (test) and the corresponding control value (control_E01, see URL) is presented. The values for male and female animals are presently in separate column headers. The ratio of the test to control value is shown, and it is from this measurement that a flag is set when it exceeds a threshold (usually 1.25=H, 0.80=L). The KOMP staff can over-ride this computer generated flag. The third column contrasts the female/male ratio measurement of the test and control groups. Then the test-ratio/control-ratio ratio (F/M-test/F/M-control) is calculated to determine if there is a deviation from the expected gender dimorphic values.

   If the abbreviation or name of the measurement is insufficient and you wish more detail about the measurement, click the measurement definition button (1) at the lower left. It brings up the definition file that relates to either µCT or histomorphometric measurements (figure 7B). Scroll up and down to find the measurement and then close the file (#5) and resume examining the summary file. At anytime you can open the previous files by clicking the navigation control buttons (#3 and #4). Similar summary files that relate to cortical and trabecular bone as determined by µCT are present when entered from the analysis page. It is from this presentation of the data that our KOMP team analyzes a study and determines whether a value or set of values is of sufficient magnitude to be scored as being either high of low. Comments are often made at his stage to convey observation that might not be apparent in the individual measurements. During the evaluation stage, the decision is made to progress to histological analysis or defer further study. When the analysis is complete, the biological code is generated based on the strength of the measurements that compose each component of the code. It is through the code generating process that the user can restrict a search based in the individual scored components.

   The subsequent tables are designed to present the quality of the data that produce the mean value in the summary table. To see the distribution of values from the individual animals in the test group for any of the mean measurement shown in the summary table, click the see detail data in the upper right corner (#2 in figure 7a). The Detail Table appears (figure 8, equivalent to the blue square in figure 1B) with a table of the individual mice listed vertically and the abbreviated measured value listed across the top of the page. The full name of the measurement can be obtained from the definition file located at the lower left (#1). In addition to the individual values, the mean, animal number, total section number, Std Dev and CV values are shown. The table represents trimmed data, which eliminates values that were deleted usually due to a technical reason and are the values used to calculate the mean of the measurements. A similar table is generated for the µCT data. To view the histological images and their computer representation, click view images (#2) located in the upper left side of the screen, which is seen in figure 9. Returning to previous screens, click the link numbered 3-5). .

   This page (figure 9, equivalent to the pink box in figure 1B) presents the values and images of the 3 histological sections from each animal that are the basis of the mean calculation for an individual mouse. The image table contains the measured values from each section from each mouse bone is presented, as well as the mean, SD and RSE. A comment field that contains any messages that the histology staff had about the quality issues or unusual findings (#1). Below the table are the 7 images that were used for the analysis. The images represent: A. Dic. The first image set containing the mineral and mineralization lines (3 layers) B. TRAP. The second image set of the TRAP stained section (2 layers) C. AP. The third image set using AP and DAPI (2 layers). D. TBlue/He. The final image of the toluidine blue stained section (1 layer) E. R-Roi. The computer rendering of the 7 fluorescent layers (A-C) and the computer defined ROI. F. P-Pseudo. Enlargement of the elements that comprise the analysis and the signals have been converted to a binary value. G. T-Trim Projection of each element onto the trabecular bone surface. The measurements from the histomorphometry are generated from this image.

   There are a number of navigation aids for this large data set. To get a rapid overview of the quality of the sections, click see DIC image (#2). It presents the mineral representation of the bone for each section and all the mice in the group. To examine individual mice, click the advance or backwards button (#3), or you can scroll up or down. Finally you can return to earlier screen with the button at the top of the screen (#5-8) Double click on any of the images and you are taken to the Image Enlargement page (figure 10A) where that image can been seen in greater detail.

   Navigate this page with the buttons on the upper left (#1) that allow you to examine scroll through all 7 images of the section set. Click back to the image table (#3) to select other image sets to examine. The image from the enlarge image page can be downloaded to your desktop by clicking the export image button (#2). It saves a jpg file, which can be opened by most photo applications. The images are shown as recorded by our imaging conditions and have been downsized in resolution to fit with the constraints of the database. The down loaded file can be adjusted (brightened) as needed (see figure 10B). The original high resolutions are maintained on our image database and are available on request.