Distinctive features

These are the parts of andes that were built to answer a specific question about trust in a result, not just to find more peptides. Each page gives the problem, how andes addresses it, a diagram, and measured numbers with their conditions.

On wording

We describe what andes does and measure it. We have not surveyed every search engine, so these pages do not claim that nothing else does something similar; where the benchmark compares with another tool, it is named and dated.

flowchart LR
    subgraph ID["Identification"]
        A["Per-file model<br/>and fragment-ion retrieval"]
        B["Chimeric cascade<br/>second peptide per spectrum"]
    end
    subgraph TR["Trust in the claim"]
        C["Knockoff variants<br/>variant FDR with a real null"]
        D["Site claims<br/>occupancy and identity"]
        E["Protein inference<br/>picked target-decoy"]
    end
    subgraph QU["Quantification"]
        F["Error-controlled<br/>match between runs"]
        G["TMT interference<br/>unmixing"]
    end
    A --> B --> C --> D --> E --> F
    A -.-> G

Feature	Question it answers	Flag
Knockoff variant FDR	Is this variant real, or a modification or isotope error that looks like one?	--knockoffs, --proteoform-fdr
Site claims and identity	Is this residue modified, and which modification is it?	sites=annotated mods
Chimeric cascade	What is the second peptide in this spectrum?	--chimeric
Retrieval and model choice	How do you search deeply without scoring every candidate?	automatic
Protein inference	Which proteins are present, at what error rate?	--protein-fdr
Match between runs with error control	Can a transferred quantity be trusted?	--lfq --mbr
TMT interference unmixing	How much of a reporter ion is not from the identified peptide?	--tmt-unmix