Overview of NAb tool
Users are prompted to provide the NAb tool with a raw instrument file produced by their luminometer, plus relevant metadata about the batch, run and samples that have been processed. Required metadata include characteristics necessary for calculations (e.g., initial dilution, dilution factor and desired curve fit method). Optional metadata include characteristics of the experiment that can be used for tracking purposes (e.g., experiment performer, incubation time and virus name). Both raw data and metadata are analyzed automatically by the system to estimate the titers at which specified levels of neutralization are seen for each sample/virus combination. Results can be integrated with other data types (such as specimen information and clinical data for study participants). The NAb tool displays neutralization curves, titers and source data in a secure, interactive, web-based interface.
Options for leveraging the NAb tool
The LabKey Server NAb tool can be accessed in two ways:
• By logging into to the Atlas Science Portal. This option is available to participating member consortia of the Global HIV Enterprise [10
• By installing and configuring a LabKey Server instance of your own. This option allows you to administer and customize your own, private LabKey Server for your lab or consortium.
Detailed information for these options is available in the "Availability and Requirements" section of this paper.
Tutorials and documentation
Full documentation and tutorials for setting up, configuring and using LabKey Server and its NAb assay are available at http://www.labkey.org.
This documentation is updated regularly to match the currently released version of LabKey Server. The LabKey Server NAb assay tutorial [14
] provides a detailed walk-through and sample data.
Assay plate options
The tool supports both low-throughput (single-plate) and high-throughput (multi-plate) NAb assays performed in either a 96-well or 384-well plate format. At present, a low-throughput assay typically utilizes a 96-well plate, usually prepared with five specimens in eight dilutions of two replicates each, as shown in Figure . A high-throughput assay typically uses 384-well plates, allowing for more samples per run. A high-throughput run may consist of up to eight plates. High-throughput samples are diluted across plates; in contrast, low-throughput samples are diluted within a single plate. Plate layout may be customized for both low- and high-throughput assays.
Figure 2 Assay dashboard. The assay dashboard provides a central location for managing different assays and their associated runs. This dashboard has been customized to include a list of all assay designs available in the current folder. It also displays a file (more ...)
Calculation of results
The NAb tool calculates the percent neutralization for each dilution or concentration after subtraction of background activity and fits a curve to the neutralization profile. It then uses this curve to calculate neutralizing antibody titers for chosen benchmarks, area-under-the-curve (AUC) and error estimates. Five-parameter, four-parameter, polynomial, and point-based methods are all currently used to calculate curve fits. For each method of curve fit, the tool calculates up to three inhibitory dilutions or inhibitory concentrations, according to the cutoff percentages chosen by the user. For example, the user might ask the tool to calculate IC20, IC50 and IC80, the concentrations at which the antibody inhibits 20%, 50% and 80% of infections. The tool also uses the fitted curve to calculate the AUC, the positive-area-under-the-curve (PostiveAUC) and curve fit error estimates. AUC is the total area under the curve, with negative regions counting against positive regions. PositiveAUC includes only the areas under the curve that are above the x-axis.
In addition to calculating curve-based neutralization titers, the tool also calculates "point-based" titers according to the method of Reed and Muench [15
]. This is done by linearly interpolating between the two replicates on either side of the target neutralization percentage.
Formal comparisons between five-parameter, four-parameter, polynomial and point-based methods for calculating titers are not currently available. In general, five-parameter, four-parameter and polynomial methods provide more averaging than point-based methods, so they generate smoother trend lines. Common techniques for calculating five-parameter, four-parameter and polynomial curve fits are described elsewhere [16
As part of ongoing development of the tool, new calculation techniques (such as a five-parameter curve fit) have been added to the tool and quickly deployed to all users. When desirable, new calculations (such as area-above-the-curve calculations) have automatically been added to results views for existing data. This meant that new insights into analysis techniques could quickly and transparently be applied to existing data, alongside old techniques. Automated and manual testing are performed before every public release of LabKey Server to ensure that analyses performed with new versions of the software produce results consistent with previous releases.
Usage scenario for the LabKey Server NAb tool
To use the NAb tool, a scientist typically follows these steps:
1. Set up an assay folder. After setting up a LabKey Server, an administrator creates an assay-type folder with appropriate permissions for user access. The home page for the folder shown in Figure includes the Assay Dashboard (the starting place for uploading and processing NAb data) and the File web part (the general-purpose tool for uploading and importing files and data to the server).
2. Configure a plate template. A user next creates a plate template to match the design of the assay, or reuses an existing template. The template maps the contents of each well, including specimen controls and replicates. Figure shows the plate template editor.
Figure 3 Plate template editor. The LabKey Server plate template editor allows you to indicate which wells contain particular specimens, replicates and controls. Processing of sample data depends on the layout of each plate, as indicated through a plate template. (more ...)
3. Create an assay design. With a plate template in hand, the user next creates an assay design that includes both this plate template and appropriate fields, such as "Sample ID" and "Initial Dilution," as shown in Figure . This design is used as the framework for uploading many individual assay runs.
Figure 4 Assay designer. The LabKey Server assay designer  provides a pre-prepared assay design for both low- and high-throughput TZM-bl NAb assays. Both the plate template and the fields for this design can be customized. This figure shows the default design (more ...)
4. Upload files and import data. After creating an assay design, a user can begin uploading files to the repository and simultaneously importing data from those files into the database. A part of this process, the assay design guides the collection of appropriate metadata for the assay run, as shown in Figure . This information is used to determine data processing and to map samples to plate locations. Information collected typically indicates whether the sample is being diluted or concentrated; the initial dilution or concentration of each sample; the dilution factor; 1-3 cut-off percentages for calculation of inhibitory concentrations or dilutions; a unique sample identifier; and the default curve fit method.
NAb assay import screen. This image shows how users indicate metadata and cut-off percentages for neutralization titers for a high-throughput NAb assay.
5. View results. Curve fits and related results are calculated automatically for each assay run during the import process. Figure shows a results view for a low-throughput NAb assay. Calculated results are displayed towards the top of the page, while raw plate data are provided at the bottom. The results page shows calculations for inhibitory concentration (or dilutions), areas under the curves and the standard deviations of controls, both viral and cell. A drop-down menu allows users to choose which curve fit mechanism is used to calculate displayed results.
Figure 6 LabKey Server NAb results view. The calculated results of a NAb run are displayed in the "Run Summary." This example shows a low-throughput NAb run that has been shared publicly on the Atlas server as part of a completed CAVD study .
Data integration and quality control
LabKey Server allows researchers to integrate information from NAb assays with other study information through sample, participant and visit identifiers [10
]. The platform also provides mechanisms for performing quality control before sharing NAb results widely and integrating these results with other data sources. Figure shows a schematic overview of how data can flow into a LabKey Server from diverse sources, undergo quality control and become available to a range of data customers through a web-based portal.
Figure 7 LabKey Server data flows. This figure shows how data flows into a LabKey Server in many forms (e.g., Excel, text and DataFax case report forms) from many sources (e.g., labs, clinics and repositories). Once within the system, data can be accessed by collaborators (more ...)
Typically, imported NAb data includes sample identifiers (sample IDs). These are used to look up the participant IDs and visit IDs associated with the samples during assay run import. Data are initially stored in a private folder, along with sample IDs, participant IDs and visit IDs. Quality control checks are then performed to identify inconsistencies in the data and users can see the errors immediately onscreen.
After the data have been reviewed privately for accuracy, they are typically then transferred to a shared integration folder called a study. A study allows groups of researchers to use participant and visit IDs to integrate data of many different types, such as specimen information, clinical data and other assay results for a study participant. Just like NAb data, all of these data types have first been reviewed for accuracy before they are transferred to a study folder for sharing and integration.
Additional features of LabKey Server
LabKey Server includes a range of additional built-in features that can be helpful to users of the NAb tool. These include a fine-grained security model, auditing, a robust API and tools for data integration. A full review of LabKey Server's features as of v10.3 is available elsewhere [10
The platform's folder- and role-based security model allows a single LabKey Server to support multiple groups privately managing NAb analyses in protected projects and folders. Data can be shared based on the role and group memberships of each user. Furthermore, finished data can be released publicly by simply lowering the permissions requirements for the particular subfolder that contains the data.
LabKey Server's assay infrastructure supports auditing and can be used to achieve GCLP. For example, imported runs can be re-imported to the system as new runs for error correction, but existing runs cannot be edited.
LabKey Server's NAb tool has been adopted by multiple consortia within the Global HIV Enterprise [10
] for uploading, processing and sharing results of the TZM-bl NAb assay. These labs use the tool as part of the Atlas Science Portal [18
], which is itself an installation of LabKey Server. The NAb tool has been used successfully by 14 labs across 5 organizations within the Enterprise: the Collaboration for AIDS Vaccine Discovery (CAVD), the Center for HIV Vaccine Immunology (CHAVI), the Vaccine Immunology Statistical Center (VISC), the HIV Vaccine Trials Network (HVTN) and the U.S. Military HIV Research Program. As of January 2011, these labs have used the NAb assay tool to upload and store over 40,000 NAb assay runs. As of May 2010, Atlas supported approximately 2,800 user accounts originating in roughly 350 organizations and 36 countries [10
Post-publication sharing of NAb assay analyses is a particularly notable aspect of how the LabKey Server NAb tool is being used. Multiple labs within the CAVD consortium have made their NAb data and analyses public on the Atlas portal [18
]. This portal provides an interactive interface that provides sortable and filterable grid views of results. Figure shows an example of a NAb assay results view available publicly on Atlas [19
]. Results from other CAVD studies can be explored using the CAVD study design browser [20