QC Summary
QC Checks
There are a number of quality control checks that are applied on a plate and sample level. See the metrics appendix page for a summary of all metrics.
Minimum SOMAmer Read Counts: Non-blank samples with less than 10 million reads will receive a FLAG for SOMAmerReads_PassFlag in the ADAT. These reads are counted in the raw counts step. Only human protein SOMAmers are part of this count, not controls. There is no specification for blank samples.
Maximum SOMAmer Read Counts: Blank samples with more than 40 million normalized reads will receive a FLAG for SOMAmerNormRead_PassFlag in the ADAT. These reads are counted in the plate scale normalization step. Only human protein SOMAmers are part of this count, not controls. There is no specification for non-blank samples. A plate where 70% of the blanks have a FLAG for this step will receive a WARNING.
Reference Correlation: This step produces a Spearman correlation coefficient describing how similar a sample is to a an external Plasma or Serum reference (see below). Blank samples with a correlation to the reference greater than 0.6 will receive a FLAG for RefCorr_PassFlag in the ADAT. A plate where 70% of the blanks have a FLAG for this step will receive a WARNING.
The reference used in this step can be found in the SOMAmer metadata, under Ref.MedNormExt.Plasma.QC or Ref.MedNormExt.Serum.QC (dependent on the input type).
QC Check: This step compares the median of each SOMAmer measurement, across the three QC sample replicates, to an external QC reference. It then calculates a SOMAmer-specific scale factor and a QC metric (QCCheckTailPercent). QCCheckTailPercent corresponds to the percentage of SOMAmers with scale factors outside the specification range (0.8–1.2). If more than 15% of the scale factors are outside of the specification range, the plate receives a FAIL.
The references used in this step can be found in the SOMAmer metadata, under Ref.QCCheck.Plasma or Ref.QCCheck.Serum (dependent on the input type).
Last updated
Was this helpful?