The Focal Adhesion Analysis Server: a web tool for analyzing focal adhesion dynamics

FA properties and visualizations

The analysis pipeline extracts and quantifies a wide range of properties. FA properties characterized in each individual image include adhesion area, marker protein intensity and the lengths of the major and minor axes. In addition to these static properties, the system also collects dynamic adhesion properties, which are quantified by recording the changes in individual adhesions between frames in the image stack. Dynamic properties currently include the FA assem­bly and disassembly rates⁶ and the focal adhesion alignment index⁴. All of these results are saved in CSV format, which is suitable for import into statisti­cal or graphing software. For users only interested in static results derived from individual images, as in an analysis of a set of fixed-cell images, all the other dynamic properties can be safely ignored.

The user is also provided with two types of visualizations that show either the entire field of view or single adhesions over time. The visualization of the entire field of view is produced for every image in the submitted image set and outlines each adhesion with a unique color (Figure 1). This visualization can be used to verify that the adhesions were correctly detected, segmented and tracked. The second visualization type shows a single adhesion segmented and tracked through time (Figure 1).

Provided that adhesions are present for at least 10 sequential images, this visualization allows the user to compare an individual FA’s properties with the appearance of the adhesion in the original image data. This suite of automatically extracted properties and visualizations enables the user to minimize the amount of laborious manual analysis normally required to quantify FA image sets.

User adjustable parameters

Several of the parameters used to analyze the images can be specified when an image set is submitted for analysis. The most important of these is the threshold used to identify the regions of the image that qualify as FAs versus background. The appropriate threshold will vary depending on the type of cell imaged and the imaging conditions. To make setting this parameter easier, we have added a feature where a single image can be submitted, segmented using various thresholds and then immediately returned to the user for visual inspection of the results obtained when the threshold is varied. The user also has the option to turn off the default watershed-based segmentation that is used to split adjacent FAs and modify the minimum or maximum FA size accepted by the system. Finally, the time between images can also be specified to ensure that the calculation of the rates of assembly or disassembly are made in the correct units.

Users have the option of providing an email address when an image set is submitted. If an email address is provided, notification of the completion of the image processing pipeline, along with a link to download the results, is sent. The system can also be used without an email address, but the user must return to the web interface to check on the status of the processing. The processing time is dependent on the number of images in the set and how many adhesions are detected during processing. Using typical input data, we tested the system throughput and found that the average processing and analysis time per image, under full load, is 13 seconds. Because the system can handle four image sets at once, we expect experimental throughput to be acceptable for everyday usage.