The Filter tool, included in the Basic Plug-in for Subio Platform, is a general-purpose tool for extracting Measurements (such as genes) that meet specified criteria. In addition to values from individual samples, it can use statistics calculated within Sample Groups, including the mean, median, standard deviation (SD), coefficient of variation (CV), and skewness. This allows you not only to filter out genes with low expression or little variation in expression, but also to select Measurements based on a wide range of criteria.
For example, you can extract genes whose expression levels exceed a specified threshold in a certain percentage of samples, or use SD or CV to identify genes with high variability across samples. Skewness can also be used to filter Measurements based on the asymmetry of their value distributions. The Filter tool is therefore useful not only for quality control before analysis, but also for selecting Measurements based on various characteristics of the data.
1. Basic Operations of the Filter Tool
Open the Filter tool and specify the target Measurement List. You can then select the Sample Groups to be analyzed individually or specify a DataSet that contains multiple Sample Groups. Next, select the Signal to be filtered and set the filtering conditions and thresholds. You can apply conditions to values from individual samples, calculate statistics such as Mean, Median, SD, CV, and Skewness within a Sample Group and apply conditions to those statistics, or specify the percentage of samples that must meet a condition. When a DataSet is selected, you can also specify how many Sample Groups must satisfy the criteria.
Because conditions can be defined as a percentage of samples, you can efficiently filter multiple groups with different sample sizes. For example, if you specify 50%, at least 3 out of 5 samples must meet the condition in a group of 5, while at least 30 out of 60 samples must meet it in a group of 60. The required number of samples is automatically adjusted according to the size of each group.
2. Adjust Filtering Conditions Using Histograms
The Filter tool displays histograms showing the distribution of the selected Signal or statistic. You can switch the histogram axes between linear and logarithmic scales. When you select Mean, Median, SD, CV, or Skewness, the histogram displays the distribution of the corresponding calculated values.
Click Preview to select the Measurements that meet the current filtering conditions. These Measurements are highlighted in black in the Viewers in the main Subio Platform window and in the histogram within the Filter tool. You can also zoom in vertically and horizontally, making it possible to examine the results even when only a small number of Measurements meet the criteria. By adjusting the thresholds and repeatedly clicking Preview, you can refine the filtering conditions while checking which Measurements are selected.
3. Save the Filtering Results
Click Extract to save the Measurements that meet the specified conditions as a Measurement List. Click Filter Out to save the Measurements that do not meet the conditions. The statistics calculated during filtering are also saved with the Measurement List, and you can examine their values in the Annotations tab.
4. Video Tutorial
The following video demonstrates the basic operations of the Filter tool. Although it was recorded using an older version of Subio Platform, the fundamental concepts remain the same, including selecting a DataSet or Sample Groups, specifying conditions based on the percentage of samples, and adjusting thresholds using visualization tools.
- 0:10 Overview of the DataSet used in the examples
- 0:25 Selecting a DataSet or Sample Groups
- 0:55 Identifying Measurements with low signals in individual samples
- 1:20 Identifying Measurements with low signals in all samples
- 1:50 Filtering based on Sample Groups
- 2:40 Filtering out Measurements with little variation between Sample Groups
- 3:10 Choosing thresholds using visualization tools
- 4:00 Example 1: Extracting genes that are not expressed
- 6:10 Example 2: Extracting genes with tightly regulated expression levels
For more information about the basic concepts of filtering, see the RNA-Seq Data Analysis Tutorial. If you prefer a shorter video demonstrating the workflow, see Short Tutorial 2: Preparing Data for Analysis Using Filters.
Related Topics
- What Are Flag Values in Microarray Data? — Explains how flag values are used to assess the quality of microarray measurements.