Protein coverage
Draws each inferred protein's sequence with the identified peptides laid under the residues they cover, marking which peptides it shares with other proteins.
What it does
After Protein inference and coverage or Bayesian protein inference runs in the Mass Spectrometry panel, this map shows the proteins it reported, best group first. Pick one to see its sequence fifty residues to a row, every residue a peptide covers shaded, and each peptide drawn as a bar under its residues. A strip above the rows shows the whole protein at once.
What this computes
- peptide placements: every occurrence of each identified peptide in the protein, by the same mapping the inference used, ambiguous residues and isoleucine/leucine included
- unique and shared peptides: a peptide found in only this protein is blue; one that also occurs in another protein of the database is orange, because it cannot say on its own which protein was present
- razor peptides: with greedy group resolution on, a shared peptide given to a better-scoring protein is drawn hollow and names the protein that took it
- the protein's score: its aggregated score, or its posterior probability after Bayesian inference
- sequence coverage: the residues inside at least one matched peptide, as a count and a percent
- per peptide: its residues, how many spectra matched it and its best score, on hover
- indistinguishable proteins: the group members the evidence cannot tell apart from the one shown
Site avoidance heat map
Draws the last site-avoidance sweep: every record against every recognition site, coloured by how far its count sits below or above the null.
Reference-gene stability
Ranks candidate reference genes by how stably they behave across your own samples, and says how many of them you actually need to normalise against.