preProcessAdat: Pre-Process an ADAT Object for Analysis

In SomaDataIO: Input/Output 'SomaScan' Data

Pre-Process an ADAT Object for Analysis

Description

Pre-process an ADAT file containing raw analyte RFU values in preparation for analysis. For more details please refer to the pre-processing how-to article

Usage

preProcessAdat(
  adat,
  filter.features = TRUE,
  filter.controls = TRUE,
  filter.qc = TRUE,
  filter.outliers = FALSE,
  data.qc = NULL,
  log.10 = FALSE,
  center.scale = FALSE
)

Arguments

adat	A soma_adat object created using read_adat(), including SeqId columns (seq.xxxxx.xx) containing raw RFU values.
filter.features	Logical. Should non-human protein features (SeqIds) be dropped? Default is TRUE.
filter.controls	Logical. Should SomaScan technical control samples be dropped? If TRUE, this retains all samples where SampleType = "Sample" (study samples) and discards all others including buffer, calibrator, and QC control samples. Default is TRUE.
filter.qc	Logical. If TRUE only samples that pass default normalization acceptance criteria will be retained. Default is TRUE.
filter.outliers	Logical. Should the adat object drop outlier samples? An outlier sample is defined by >= 5% of filtered SeqIds exceeding +/- 6 MAD and 5x fold-change from the median signal. This filter is typically appropriate for studies on plasma, serum, and other biological matrices generally exhibiting homeostatic characteristics. For studies on matrices such as tissue homogenate, cell culture, or study designs containing client-provided background lysis buffer controls (or similar), this filter will likely not be appropriate. Default is FALSE. If set to TRUE it is highly recommended that filter.controls is also set to TRUE
data.qc	Character. Character vector of variable names for which data QC plots are desired. Default is NULL, which does not generate any QC plots. Note: These plots are for visual inspection only, no samples or features are dropped from the output soma_adat object.
log.10	Logical. Should the RFU values be log10 transformed? Default is FALSE.
center.scale	Logical. Should the RFU values be Z-transformed (centered and scaled)? Default is FALSE. If set to set to TRUE it is highly recommended that log.10 is also set to TRUE

Details

The soma_adat object is pre-processed with the following steps:

1. Filter features -> down to human protein analytes

2. Filter samples -> by the following order and criteria: a) Retain study samples only (dropping buffer, calibrator, and QC samples) b) Only those that pass default normalization acceptance criteria c) Those not identified as outliers.

3. Data QC -> plots of normalization scale factors by clinical covariates

4. Transformations -> log10, center, and scale analyte RFU values

Value

A soma_adat object.

Author(s)

Caleb Scheidel

Examples

preProcessAdat(example_data, data.qc = c("Age", "Sex"))

SomaDataIO documentation built on June 8, 2025, 10:13 a.m.