• 3D images of 90 clones of randomly CD-tagged 3T3.
  
• 2D images of 175 clones of randomly CD-tagged 3T3, for p-value thresholds used to select features of 0, 0.1, and 0.2.
  
• 3D images of wildtype and 11 mutants of GFP-tagged UCE in HeLa cells.

PSLID collects and structures 2-D through 5-D fluorescence microscope images, annotations, and derived features in a relational schema. It is designed so that interpretations as well as annotations can be queried. The annotations in PSLID, composed of 44 linked tables with publicly available descriptions, provide a thorough description of sample preparation and fluorescence microscope imaging.

Image interpretation is achieved using Subcellular Location Features that have been shown to be capable of recognizing all major subcellular structures and of resolving patterns that cannot be distinguished by eye.

The fundamental unit of PSLID is an image set, which is simply a logical grouping of images. Image sets can be defined at the time of image loading, or they can be defined by searching for images that meet specified criteria (e.g., all images of "actin" or all images that are similar to a query image). They can also be created by analysis functions such as cluster analysis (e.g., the images in each cluster found by cluster analysis can be put into distinct sets).

Analysis capabilities that are incorporated in PSLID include:

• Searching for images by context (annotations) or content
• Ranking images by typicality within a set: e.g., to choose an image for presentation or publication
• Ranking images by similarity to one or more query images: "searching by image content" or "relevance feedback"
• Comparing two sets of images (hypothesis testing): e.g., to determine whether a drug alters the distribution of a tagged protein
• Training a classifier to recognize subcellular patterns
• Using a trained classifier to assign images to pattern classes: e.g., assigning images to "positive" or "negative"
• Clustering images by their subcellular patterns: e.g., finding "subcellular location families" within a large set of images