User Tools

Site Tools


prolineconcepts:spectralcountdef

This is an old revision of the document!


Compare Identification Result with Spectral Count

Definition

  • The peptide spectral count consist in counting the number of spectra which matches the current peptide. Thus, it's equal to the number of pepitde spectrum matches (PSM).
  • Protein basic spectral count (BSC) is equal to the sum of the peptide spectral count for all peptides which identify the protein.
  • Protein specific spectral count (SSC) is equal to the sum of the peptide spectral count for specific peptide only. A specific peptide, is a peptide which does not identify any other protein (or more precisely protein in other protein sets) in the context of the identification result.
  • Protein weigthed spectral count (WSC) is the Protein specific spectral count + sharing-weighted spectral count of shared peptide.
Example calculation of spectral count

Specificity and weigth reference

The peptide specificity and the spectral count weight could be defined in the context of the Identification Result where the spectral count is calculated as shown in previous schema. It could also could be done using an other Identification Result as reference, like using the common parent Identification Result. This allow to considere only identified and validated protein in the merge context.

If we considere the following case, where Sample1 Identification Result is the merge of Replicat1 and Replicat2.

If the spectral count calcultation is done at each child level, aligning protein sets identified in parent to protein sets in child, we get the following result :

Sample1 ProteinSets Replicat1 Replicat2
Ref Prot. BSC SSC WSC Ref Prot. BSC SSC WSC
P2 P2 5 2 3.5 P3 7 7 7
P3 P3 4 1 2.5 P3 7 7 7

We can see that when different parent protein sets are seen as one protein set in a child, the spectral count is biased. Now, if we aligne on child protein rather then protein sets, we get the following result :

Sample1 ProteinSets Replicat1 Replicat2
Ref Prot. BSC SSC WSC Ref Prot. BSC SSC WSC
P2 P2 5 2 3.5 P2 2 0 0
P3 P3 4 1 2.5 P3 7 7 7

Again, when considering specificity at protein set level, the result of spectral count in Replicat2 is not representative, as it has a null SSC and WSC !

A way to make some small correction is to define the specificity of the peptide and their weight at the parent level, and apply it at the child level. Therefore, specific peptides for P2 is pe8 and for P3 it is pe6 and pe7. For peptide weight, if we consider pe4 for example, it will be define as follow :

  • Weight pe4 for P2 = 1/3 ⇒ P2 has 1 specific peptide for a total of 3 (if we considere P2 and P3 which are proteins identified by pe4)
  • Weight pe4 for P3 = 2/3 ⇒ P3 has 2 specific peptide for a total of 3 …

The spectal count result will thus be :

Sample1 ProteinSets Replicat1 Replicat2
Ref Prot. BSC SSC WSC Ref Prot. BSC SSC WSC
P2 P2 5 2 3 P2 2 0 1
P3 P3 4 1 3 P3 7 5 6.34

In Proline

Actually, spectral count is calculated for a set of hierarchy related Identification Results. In other words, this means that Identification Results should have a common parent.

Firstly, the peptide spectral count is calculated using following rules

  • Equal to Peptide Spectrum Matches Count if Identification Result is a leaf (not issued from a merge)
  • Sum of child peptide spectral count in case of identification resulr merge
  • Sum of validated child peptide spectral count in case of Search Result merge. Validated child PSMs are PSMs of child Search Result which satisfy validation applied to parent Identification Results.

Once, peptide spectral count is known for each peptide, protein spectral count is calculated using followig rules

  • Protein BSC = sum of peptide spectral count
  • Protein SSC = sum of peptide spectral count for specific peptide only
  • Protein WSC = SSC + weighted peptide spectral count for shared peptides. The weight of a peptide for a given protein (P1) = peptide SC x ( number of specific peptides of P1 /number of peptides specific peptides of all protein identified by the peptide)
prolineconcepts/spectralcountdef.1412323096.txt.gz · Last modified: 2014/10/03 09:58 by 132.168.72.225