Tricky cases

Beads no attribute explains, user-defined attributes, and the parsimony rule that arbitrates.

Chapter 4 — Tricky cases & verdict

4.1 · Non-explained beads

Sometimes masking strips a positive bead of all its candidate attributes. SAgA then labels it a non-explained bead and returns no antibody for it. There are two ways to read that:

A non-explained bead with its two possible interpretations — A positive bead that has lost every candidate attribute. SAgA returns nothing for it, and leaves the interpretation to you.

A false positive of the assay. Some beads are known to give frequent non-specific binding; here that is the most probable explanation.
An incomplete attribute list. A relevant target may simply be missing from the bead's candidates — which the next section turns into a fix.

4.2 · User-defined attributes

When the standard attributes cannot explain a bead but the literature supports a specific target, you can declare a user-defined attribute. SAgA will take it into account like any other attribute.

All standard attributes are masked by negative beads, leaving the positive bead non-explained — The standard alpha and beta attributes are all carried by negative beads, so they are struck from the positive bead too. With nothing left, it would be flagged as non-explained.

A user-defined QB2A5 epitope rescues the bead; QB2A5 = 4000 is returned — The user defines **QB2A5** (an alpha-beta attribute — the serogroup of the DQ2.5 combination, DQA1*05 + DQB1*02). No single negative bead carries both together, so QB2A5 is unique to the positive bead — its 4000 MFI is assigned to QB2A5 and reported, consistent with the literature.

4.3 · Sparsity

Custom attributes can also let SAgA explain the same signal with fewer antibodies. When two interpretations both fit the data, SAgA is sparse: it returns the solution with the fewest attributes. Of course, you should only add a new attribute when there is scientific justification for creating it.

Three DQA1 alpha chains defined independently yield three separate antibodies at 3000 — Three beads carry three different DQA1 alpha chains (DQA1*04, *05, *06), all reacting at 3000. Defined independently, they yield three separate antibodies.

Defined as a single shared QA456 epitope, the three collapse into one antibody at 3000 — Defined as a single shared epitope **QA456**, the three collapse into one. Given the choice, SAgA reports a single QA456 antibody at 3000 — the sparser solution.

Cases for this part