Co-IP-MS can support a protein interaction network only after the identification table is filtered. The raw list from a bait IP is a digest inventory. It is not a set of edges. Antibody chains, bead binders, and abundant lysate proteins appear beside true neighbors. Drawing every identified name around the bait converts background into topology. Network construction starts with background removal, replicate consistency, and a stated confidence rule. Public databases can annotate that filtered set. They should not import extra neighbors that the current IP never recovered. Visualization then displays ranked candidates. It does not certify the figure as a finished interactome.

Researchers who already have bait IPs, matched negative IPs, and a written network claim can review MtoZ Biolabs IP-Mass Spectrometry Service for feasibility. Pair-level method choice after a short candidate list is covered in Measuring Protein-Protein Interactions.

An Identification List Is Not Yet a Protein Interaction Network

A Co-IP-MS search reports proteins present in the digest. MaxQuant against UniProt or a client FASTA assigns accessions and intensities. The table answers presence under the submitted wash. It does not assign a binary contact to every pair of names.

Published AP-MS maps that later appear as networks were built from scored enrichments, not from unfiltered protein lists (Huttlin et al. 2015). Co-enrichment can still be indirect (Hein et al. 2015). A bridging subunit, a shared organelle contaminant, and a true neighbor can occupy the same table.

Do not treat identification depth as network size. Co-IP-MS typically reports hundreds to thousands of proteins, and the count is sample-dependent. After filters, the usable graph is the bait-centered subset that survives controls and replicates.

If the bait is missing from the bait IP table, stop. No network should be drawn from a capture that did not recover the intended protein. Product-level digest choices for already-run IPs are covered in Cellular IP Product Mass Spectrometry Detection.

The comparison below keeps an unfiltered identification list, a control-aware candidate set, and a ranked display on separate tracks.

Figure 1. Unfiltered IDs vs Filtered Candidates vs Ranked Network Display.

Background Removal Before Any Edge Is Drawn

Background removal is the first network filter. The negative IP must use the same input, beads, wash, digest, and LC-MS/MS method. Isotype IgG or bead-only tubes address reagent binding. Empty-tag or bait-depleted inputs address construct artifacts.

Proteins equally present in the control IP are not edges. They remain annotated in the identification table. Antibody chains, keratins, and trypsin are expected reagent or handling proteins. They should be labeled, not plotted as novel partners.

Recurrent-contaminant catalogs such as the CRAPome help interpret leftovers after experimental subtraction (Mellacheruvu et al. 2013). A CRAPome flag is not an automatic delete key. Some frequent names can still be real neighbors in a specific complex. The current control intensity decides whether the name stays a candidate.

Do not replace the matched control with a published contaminant list. Do not subtract names only because they look abundant. Relative intensity versus the control IP is the usable comparison.

Replicate Consistency as a Partner Filter

Replicates test whether a candidate reappears when the capture is repeated. Biological replicates address culture or tissue variation. Technical repeats of the same eluate address run noise. The two are not interchangeable.

A partner that appears in one bait IP and vanishes in the matched repeats is a weak node. Keep it off the first network figure. A partner that is enriched versus control in every biological replicate can enter ranking.

Replicates do not replace the negative IP. Three bait IPs without a control still describe repeated background. Three control IPs without bait recovery still describe reagent binding.

Write the consistency rule before search. Common practice is to require detection and enrichment in a stated fraction of biological replicates. The rule should be the same for every bait in a multi-bait study.

Interaction Confidence After Enrichment

Confidence for an experimental edge is enrichment confidence, not database popularity. Scoring such as SAINT ranks bait IPs against a control set when replicates exist (Choi et al. 2011). Fold change or intensity ratio versus the matched control is a simpler alternative when the control design is clean.

State the score type. A SAINT probability, a control-normalized intensity ratio, and a spectral-count difference are not the same number. Do not mix them on one color scale without a legend.

High confidence still means co-enriched under this wash. It does not mean direct binding. Orthogonal pair tests, such as a defined-pair pull-down, remain a later step when a named pair must be examined outside the original lysate. That follow-up can be scoped through the GST Pull Down Assay Service.

End-to-end capture-to-search logic that feeds this ranking is summarized in How to Combine Co-IP with Mass Spectrometry for Protein Interaction Analysis.

Database Integration and Visualization Without Invented Edges

STRING, BioGRID, and similar resources can annotate a filtered experimental set. STRING aggregates experiments, text mining, and interolog transfer into a composite score (Szklarczyk et al. 2019). BioGRID collects published experimental reports. Neither resource measures the current IP.

Use the database as context. A prior edge can justify keeping a borderline experimental name for discussion. A pathway annotation can group nodes after ranking. Reverse the order, and imported neighbors become fake experimental edges.

Do not add STRING first neighbors that this IP never identified. Do not hide node origin. Experimental nodes, database-only nodes, and reagent proteins should use different symbols if they appear on one canvas.

Visualization is a display of the ranked table. Node size can encode bait-normalized intensity. Edge weight can encode enrichment score. Color can encode replicate support. A network viewer can render that table. The figure remains a candidate map. Mixed enrichment-MS projects that include several baits can also be reviewed under the MS-Based Protein-Protein Interaction Analysis Service.

Network object	What it can support	What it cannot support
Raw ID in the bait IP	Presence in that digest	An interaction edge
Enrichment vs control IP	Candidate partnership	Direct binding
Replicate overlap	Stability of the candidate	A substitute for the negative IP
SAINT or intensity rank	Priority for follow-up	A finished interactome
STRING or BioGRID overlay	Prior or transferred context	A new experimental edge

The table is a construction rule, not a ranking of software.

Samples, Deliverables, and a Network-Ready Intake

For a full Co-IP-MS workflow that will later support a network figure, the usual starting point is a commercial IP-grade antibody supplied by the client, plus cells or tissue that express the bait. Vector construction, overexpression, and knockout are outside this intake. For already-run IPs, send visible gel bands, washed beads, or liquid products. Planning amounts for liquid IP products are at least 20 ug total protein at more than 0.5 ug/uL, with SDS kept below 0.1 percent. High salt, undepleted detergent, and severely degraded material should not be shipped.

State the bait, control set, replicate plan, and the sentence the network figure is allowed to support. State whether STRING or another database will be used only as annotation.

Orbitrap-class high-resolution LC-MS/MS is the usual acquisition platform. Search should use a stated protein database in MaxQuant or an equivalent disclosed workflow. Relative intensity can be compared across bait and control IPs.

Intake is faster when filter rules are mapped before shipment. The diagram below links those checks to a network-ready request.

Figure 2. Matching Filter Checks With Network Construction.

Deliverables typically include identification tables, a control-filtered candidate list, a replicate-consistency note, and an optional network display of that ranked set. Deliverables do not include a claim of direct binding, a guaranteed node count, or a certificate that imported database edges were measured here. To start intake, send the bait, control and replicate plan, and the intended network claim to info@MtoZ-Biolabs.com or use the site inquiry form so filter rules can be confirmed before the figure is scoped.

Frequently Asked Questions

1. How many biological replicates are needed before drawing a Co-IP-MS network figure?

Enough to apply the written consistency rule. Two or three biological bait IPs plus matched controls are a common planning range for ranking. A single bait IP can produce an identification table. It cannot support a stability-filtered network.

2. Can STRING neighbors be added as extra nodes if they were absent from this IP?

They can appear in a discussion schematic if they are marked as database-only. They should not be drawn as experimental edges. The experimental figure should stay limited to proteins recovered and enriched in this study.

3. Should ribosomal proteins be deleted automatically before visualization?

No. Ribosomal proteins are frequent background, but some baits sit in translation-related complexes. Compare control intensity first. Delete or keep by enrichment, not by protein class.

4. Does a high SAINT score replace reciprocal IP?

No. SAINT ranks enrichment versus controls in the current design. Reciprocal IP tests whether the candidate can recover the original bait. The two answers are different.

5. What should a feasibility email attach when the request is a network figure?

Attach bait and antibody information, the control and replicate plan, the intended claim sentence, and whether a public database will be used only as overlay. Feasibility review will treat the figure as a display of a filtered table.

Researchers can convert a filtered identification table into a control-aware candidate network, then confirm bait, controls, replicates, and display rules through MtoZ Biolabs IP-Mass Spectrometry Service intake.

Reference

1. E.L. Huttlin, L. Ting, R.J. Bruckner, F. Gebreab, M.P. Gygi, J. Szpyt, S. Tam, G. Zarraga, G. Colby, K. Baltier, et al. (2015). The BioPlex Network: A Systematic Exploration of the Human Interactome. Cell, 162, 425-440. https://doi.org/10.1016/j.cell.2015.06.043
2. M.Y. Hein, N.C. Hubner, I. Poser, J. Cox, N. Nagaraj, Y. Toyoda, I.A. Gak, I. Weisswange, J. Mansfeld, F. Buchholz, et al. (2015). A human interactome in three quantitative dimensions organized by stoichiometries and abundances. Cell, 163, 712-723. https://doi.org/10.1016/j.cell.2015.09.053
3. D. Mellacheruvu, Z. Wright, A.L. Couzens, J.P. Lambert, N.A. St-Denis, T. Li, Y.V. Miteva, S. Hauri, M.E. Sardiu, T.Y. Low, et al. (2013). The CRAPome: a contaminant repository for affinity purification-mass spectrometry data. Nat. Methods, 10, 730-736. https://doi.org/10.1038/nmeth.2557
4. H. Choi, B. Larsen, Z.Y. Lin, A. Breitkreutz, D. Mellacheruvu, D. Fermin, Z.S. Qin, M. Tyers, A.C. Gingras, A.I. Nesvizhskii (2011). SAINT: probabilistic scoring of affinity purification-mass spectrometry data. Nat. Methods, 8, 70-73. https://doi.org/10.1038/nmeth.1541
5. D. Szklarczyk, A.L. Gable, D. Lyon, A. Junge, S. Wyder, J. Huerta-Cepas, M. Simonovic, N.T. Doncheva, J.H. Morris, P. Bork, et al. (2019). STRING v11: protein-interaction networks, with increased coverage and integration. Nucleic Acids Res., 47, D607-D613. https://doi.org/10.1093/nar/gky1131