PCA Outlier Screen

Flag PCA samples outside a caller-selected radius.

Description

A radial PCA screen measures distance from a supplied center in a two-component plane.

A radial PCA screen measures distance from a supplied center in a two-component plane. It differs from calculating PCA or modeling an elliptical covariance contour.^[1]

Flag PCA samples outside a caller-selected radius.

When to use PCA Outlier Screen

Use this tool to flag PCA samples outside a caller-selected radius.

Prepare the upstream counts, expression values, sequences or record lists requested by this tool. Keep corresponding features in the same order and confirm count units, transformations and caller-selected criteria.

How to use PCA Outlier Screen and understand its results

Prepare the upstream counts, expression values, sequences or record lists requested by this tool. Keep corresponding features in the same order and confirm count units, transformations and caller-selected criteria.
Enter sampleIds, pcOne, pcTwo, centerPcOne, centerPcTwo, radius using the field table and calculation rules, then run the tool.
Read outlierIds, distances. The explanation below defines the units, signs, ordering and any flags or placeholder values.
If the tool reports an input issue, correct the values and run it again. Apply the stated assumptions when interpreting a successfully returned result.

Input field	Meaning	Value type; see rules below	Requirement
sampleIds	Sample ids	list of string	Required
pcOne	Pc one	list of number	Required
pcTwo	Pc two	list of number	Required
centerPcOne	Center pc one	number	Required
centerPcTwo	Center pc two	number	Required
radius	Radius	number	Required

Enter aligned nonempty IDs and two coordinate lists, a finite center and nonnegative radius. Return every distance and IDs strictly outside the radius. Points on the boundary are retained; zero radius accepts only the center.

distance_i=hypot(PC1_i−center1,PC2_i−center2); outlier if distance>radius

Units, indexing, boundary conditions and exceptional cases are defined in the calculation rules above.

Output field	Meaning	Value type
outlierIds	Outlier ids	list of string
distances	Distances	list of number

json

{
  "input": {
    "sampleIds": [
      "a",
      "b",
      "c"
    ],
    "pcOne": [
      0,
      1,
      10
    ],
    "pcTwo": [
      0,
      1,
      10
    ],
    "centerPcOne": 0,
    "centerPcTwo": 0,
    "radius": 5
  },
  "output": {
    "outlierIds": [
      "c"
    ],
    "distances": [
      0,
      1.4142135623730951,
      14.142135623730951
    ]
  }
}

Limitations and assumptions

No PCA fit, variance scaling, covariance weighting or outlier probability is computed. Component units, center and radius must match the intended upstream representation.

Numeric inputs must be finite. Follow the calculation-specific rules for overflow, underflow and exceptional cases. Lists are limited to 100,000 entries; the rules above may require smaller limits. Missing required fields, extra fields, wrong value types and violations of the calculation rules are input errors.

Alternative or Complementary methods

Use Silhouette Score: Compute a silhouette width from within- and between-cluster distances. For a complementary task, use Correlation Distance Matrix: Build a Pearson correlation-distance matrix for hierarchical clustering.

References

PCA Outlier Screen — tool.io

Similar or alternative tools

Silhouette Score
Compute a silhouette width from within- and between-cluster distances.
Correlation Distance Matrix
Build a Pearson correlation-distance matrix for hierarchical clustering.