Annotating Molecular Interactions in MINT

http://mint.bio.uniroma2.it/mint/Welcome.do

ANNOTATING PROTEIN INTERACTIONS:

STANDARDS AND BASIC ASPECTS

What is archived in MINT?

• MINT collects P-P interactions experimentally verified

• The interaction representation is according to PSI-MIs

ontology and nomenclature

• The interaction types collected in MINT are:1. Colocalizations2. Physical interactions3. Direct interactions (enzymatic reactions)

The PSI Molecular Interaction work group develops a common data standard that allows user to retrieve all relevant data from different sites and perform comparative analysis of different datasets

Curation standards

Interaction type

What is critical in the annotation process?

•Organism identification

•Proper Uniprot_ID detection

•Experimental features annotations

•Participant features annotations

http://www.nature.com/nbt/consult/index.html

Papers selection

•Title

•Abstract

•Materials and methods

•Quick look at the figures

THE SECOND BIOCREATIVE CHALLENGE:DATASETS PREPARATION

MINT curation projects

•Journals curation (FEBS letters, EMBO Journal, EMBO Reports)

•Thematic curation (Domains, VirHostome)

DELIVERED DATASETS

•TRAINING SET(128 pmids 1182 interactions)

•TEST SET(221 pmids 1489 interactions)

Biocreative training-set

•The MINT/BioCreative positive test set is composed of papers extracted from volumes of FEBS letters, EMBO Journal and EMBO Reports published from january to july 2006

•All the papers not curated from the above mentioned volumes are considered belonging to the negative test set

•For each interaction is reported the best interaction description sentence coming out from both the body text and the figure legends

ANNOTATION AND PREDICTION:USE CASES

Biocreative annotation flaws: examples

•FALSE POSITIVESPhosphorylation

•FALSE NEGATIVESPositive controls

•MISINFORMATIONComplexes

Use cases

In situ colocalization of NM1 and Pol IImmunostaining of isolated nucleoli showed that actin and NM1were enriched in nucleoli and their distribution correlated withupstream binding factor (UBF) and fibrillarin (Fig 1A). NMIcolocalized with Pol I and green fluorescent protein–RPA53, aprotein that decorates the active subpopulation of Pol I (Fig 1B).The colocalization between Pol I and NM1 was furthercorroborated by immunogold electron microscopy using intactHeLa cells (supplementary Fig 1 online). NM1 was distributedthroughout the nucleoplasm but was largely excluded from thenucleoli, except the fibrillar centres. These nucleolar subcompartmentsare known to contain Pol I and Pol I-specific transcriptionfactors (Scheer & Benavente, 1990; Dundr et al, 2002). Inaddition, both a novel autoimmune serum against Pol I (S57299;see supplementary Figs 2,3 online) and a peptide-specific affinitypurifiedantibody against NM1 blocked Pol I transcription in vitro(Fig 1C,D). These data support previous studies demonstrating akey role for NM1 in Pol I transcription (Fomproix & Percipalle,2004; Philimonenko et al, 2004).

1. Pol I is a complex2. No sentences are available for the illustrated colocalizations

This experiment describes a genetic interaction

Use cases

Post-translational modification or PPI?

Use cases