Dear Wei
From the peptide sequence data that you submitted for protein identity assignment there were a number of peptides that would match the sequence of some proteins. If you are lucky the matched peptide may be a unique segment of the protein and it would increase the probability that the peptide sequenced originates from the chosen protein.
If there are several peptide sequences that match the same protein but in different parts of the protein these are indicated in red in your figure.
The number of amino acids in the correct sequence in the different peptides that are positively identified in the protein relative to the total number of amino acids in the protein would give you the percentage coverage. In the sample you provided only 8% of the correct sequence of the peptides were matched to the amino acid sequence of the protein identified as a possible hit. The greater the coverage the more certain you can be about the protein identity being the correct protein, although several proteins of a class may share homology making it difficult to make the correct decision about the protein identity. This is why you need to specify possible modifications and your acceptable false discovery rate. Longer peptides give a better chance of correct identity while peptides of less than 5 amino acids could occur in almost any protein so any protein could be a possible hit.
Hope this gives you some idea of what the coverage is about.
Sequence coverage is simple math, the number of amino acids in a specific protein sequence that were found in the peptides sequenced in your MS/MS study. Duncan Cromarty above went into the confidence that you might have of those peptide to protein assignments. In general getting above 30% in sequence coverage is nearly impossible in most LC-MS/MS studies. The number of ambiguous peptides (i.e., those that might match more than one protein sequence) is larger, the shorter the peptides sequenced, and is usually factored into the "score". There was a raging debate in the proteomics community about a decade ago over how many unique peptides were needed to truly identify the source protein for those peptides. Since publications and grant renewals depended on the answer, many wanted a single unique peptide, others argued two, and still others argued three or more. Jenny van Eck presented an LC-MS/MS study of human samples at an AACC meeting where she reported all the 1, 2, and 3 peptide proteins identified by the study. I don't remember the exact numbers, but each additional peptide required for protein ID reduced the number of proteins identified by an order-of-magnitude. By the time you got to a requirement of 3, there were only a dozen or so proteins positively identified from almost 1500 for a 1 peptide call. No one liked that result, since it threatened their academic livelihoods, so now the debate is over False Discovery Rate and it's intrinsic meaning for 1 peptide calls.
Halligan and Dratz et al., (2004) presented a theoretical study of protein sequence tag length needed to make an unambiguous protein call. Their conclusion was that some 10% of protein calls would never be unambiguous independent of the tryptic peptide sequence length and that even this was highly depended on the ppm accuracy of the MS (< 1 ppm, required).
- Badges
- Science topic
- Similar topics
- Filing