Classification schema

One of the analysis steps performed by ProteinClusterQuant is the analysis of all protein pairs in the constructed protein-peptide network. A protein pair is formed by two protein nodes sharing a peptide node, and that may have other peptide nodes unique for each one of them. This analysis consists on the systematic detection of peptide node ratio outliers in each one of the protein pairs analyzed. The outlier detection can be performed by two different ways, corresponding to two different classification schema as described below. Both classification schema are able to handle peptide nodes having infinity ratio values (only detected in one of the samples), but do not further consider protein pairs of which of the peptide nodes has not been quantified. Each of the classification schema then further classify all protein pairs according to the presence of peptides nodes with a significantly different ratio value, either being the shared peptide node or one (or both) of the unique peptide nodes. Alternatively, none of the peptide nodes may differ significantly.
In case a significantly different peptide node is present, protein pairs are further distinguished based on the peptide node’s connectivity: one edge (unique) or two edges (shared) or three or more edges (shared by a third or more protein nodes). A connectivity of more than two edges can explain the shared peptide node’s significance within the protein pair analyzed. If only two edges are identified the ratio measurement of the shared peptide node remains unexplained. Thus it is either an incorrect measurement or caused by an additional protein edge not identified in the experiment and thus remains unexplained.

When analyzing protein pairs, two different schemata were used:

• Classification schema 1
• Classification schema 2

Classification schema 1:

• The classification schema 1 separates out all of the possible outcomes of protein pairing, depending on having actual ratio values, mixed infinities and ratio values or all infinity values. In case of having actual ratio values in all three peptide nodes, it applies a statistical test to detect a potential outlier.

Classification code	Description	Is inconsistent?
0	Having actual ratio values but not significantly different.	FALSE
1	All are infinities, same sign. No ratios.	TRUE
2	Mix of actual ratio values and infinities.	TRUE
3	One or both unique peptide nodes are significantly different.	TRUE
4	Not classified. It may have a mix of infinities (with different sign).	FALSE
5	Not classified. Some missing value (not quantified).	FALSE
6	Shared peptide node is significantly different and it may be explained by a third protein node.	TRUE
7	Shared peptide node is significantly different and it is not shared by a third protein node.	TRUE

Classification schema 2:

• The classification schema 2 is based on a user-defined cutoff threshold which is applied to the differences between the peptide node ratios. Depending on which peptide nodes are significantly different, the protein pair is classified accordingly.

Classification code	Description	Is inconsistent?
1	Shared peptide node is significantly different and is not shared by other protein node.	FALSE
2	Shared peptide node is significantly different and is shared by other protein node.	TRUE
3	Unique peptide node is significantly different, or the difference between the two unique peptide nodes is larger than threshold.	TRUE
4	Both unique peptide nodes are significantly different and shared peptide node is significantly different in between.	TRUE
5	Internal error (For internal error control of the software).	FALSE
6	There is not a significant difference. Difference between unique peptide nodes is less than threshold.	FALSE
7	Not classified. There is not enough data, i.e. some of the unique peptide nodes is missing.	TRUE