Which peptide notation does it read?

ProForma. Modifications can be written as masses ([+229.162932]), Unimod names ([TMT6plex], M[Oxidation]) or accessions ([UNIMOD:35]), and every one of the 1,560 Unimod records resolves.

What is the difference between the two protein inference methods?

Protein inference and coverage scores each protein by a rule over its peptides' scores, as OpenMS's basic algorithm does. Bayesian inference (Epifany) instead fits a probabilistic model in which proteins are present or absent and peptides are emitted by present proteins, and reports the posterior probability of each protein, each group and each match.

What does it need for Bayesian inference?

Matches scored as posterior error probabilities or posterior probabilities, the database they were searched against, and three model parameters: the protein prior, the probability that a present protein yields a peptide, and the probability that a peptide is reported with no present parent.

How are shared peptides handled?

Your choice: counted for every protein they map to, given to their best protein only (razor peptides), resolved per connected set of proteins, or explained by the Bayesian model.

Is anything uploaded?

No. Every calculation runs in the browser on the files in your project.

Analyze

Mass spectrometry: peptide masses, b ladders, isotopes, spectrum scoring, q-values, protein inference

Weigh a peptide, generate its fragment ladder and isotope envelope, score a spectrum against it, estimate the false discovery rate of a search, and infer which proteins are present. Peptides are written as ProForma with Unimod modifications, spectra come straight from an MGF run file, and protein inference reads a table of peptide-spectrum matches and the FASTA database they were searched against: by score rules with razor peptides, or by Bayesian inference that reports a posterior probability for every protein. Every number is checked against pyteomics or OpenMS over a whole public dataset.

Open Mass spectrometry: peptide masses, b ladders, isotopes, spectrum scoring, q-values, protein inference Open the Mass spectrometry sample

The 14 best-evidenced peptides of the first ProteomeXchange dataset (PXD000001), for peptide masses, fragment ladders, isotope envelopes and their 71 annotated spectra

What it does

01Monoisotopic and average peptide mass, and m/z at every charge, for ProForma peptidoforms with Unimod names or accessions
02a, b, c, x, y and z fragment ladders with neutral losses, and the most probable isotopic composition
03Coarse isotope envelopes: the M, M+1, M+2 peaks with their masses, m/z and relative abundances
04HyperScore of a spectrum against a peptide, in Da or ppm, with spectra read straight from an MGF run file
05Target-decoy q-values and FDR, with the +1 and finite-sample corrections of Levitsky et al.
06Protein inference from a PSM table and its FASTA database: best, sum or product of peptide scores, PEP scores as posteriors, indistinguishable groups
07Shared peptides settled per peptide (razor) or per connected set of proteins
08Bayesian protein inference (Epifany): a posterior probability for every protein and group, and each match's posterior after inference, with the protein prior and emission probabilities chosen by OpenMS's own grid search over a target-decoy run
09Ambiguous database residues (B, J, Z, X) and isoleucine = leucine when mapping peptides to proteins
10Sequence coverage drawn on each protein, and a theoretical 2-D map of every reported protein by pI and mass

What it runs on

Faithful ports of pyteomics and OpenMS 3.5.0, each checked against the real library over every one of the 2,382 identifications and 6,103 spectra of PRIDE dataset PXD000001 and its own 4,499-protein search database; element masses and isotope abundances are NIST's and modifications come from Unimod. The panel runs on @ports/massspec, @ports/protein, @ports/workflow.

Everything runs client-side. Files are parsed in the browser and projects are saved on this machine, so an unpublished sequence is never uploaded.

Questions

Which peptide notation does it read?: ProForma. Modifications can be written as masses ([+229.162932]), Unimod names ([TMT6plex], M[Oxidation]) or accessions ([UNIMOD:35]), and every one of the 1,560 Unimod records resolves.
What is the difference between the two protein inference methods?: Protein inference and coverage scores each protein by a rule over its peptides' scores, as OpenMS's basic algorithm does. Bayesian inference (Epifany) instead fits a probabilistic model in which proteins are present or absent and peptides are emitted by present proteins, and reports the posterior probability of each protein, each group and each match.
What does it need for Bayesian inference?: Matches scored as posterior error probabilities or posterior probabilities, the database they were searched against, and three model parameters: the protein prior, the probability that a present protein yields a peptide, and the probability that a peptide is reported with no present parent.
How are shared peptides handled?: Your choice: counted for every protein they map to, given to their best protein only (razor peptides), resolved per connected set of proteins, or explained by the Bayesian model.
Is anything uploaded?: No. Every calculation runs in the browser on the files in your project.

Also searched as

peptide mass calculator · peptide m/z calculator · fragment ion calculator b and y ions · isotope envelope calculator · HyperScore spectrum scoring · target-decoy FDR and q-values · protein inference from PSMs · razor peptides · Epifany Bayesian protein inference · protein sequence coverage from peptides · MGF spectrum viewer · TMT peptide mass · Unimod modification masses