The CAFA challenge reports improved protein function prediction and new functional annotations for hundreds of genes through experimental screens

Abstract

BackgroundThe Critical Assessment of Functional Annotation (CAFA) is an ongoing, global, community-driven effort to evaluate and improve the computational annotation of protein function.ResultsHere, we report on the results of the third CAFA challenge, CAFA3, that featured an expanded analysis over the previous CAFA rounds, both in terms of volume of data analyzed and the types of analysis performed. In a novel and major new development, computational predictions and assessment goals drove some of the experimental assays, resulting in new functional annotations for more than 1000 genes. Specifically, we performed experimental whole-genome mutation screening in Candida albicans and Pseudomonas aureginosa genomes, which provided us with genome-wide experimental data for genes associated with biofilm formation and motility. We further performed targeted assays on selected genes in Drosophila melanogaster, which we suspected of being involved in long-term memory.ConclusionWe conclude that while predictions of the molecular function and biological process annotations have slightly improved over time, those of the cellular component have not. Term-centric prediction of experimental annotations remains equally challenging; although the performance of the top methods is significantly better than the expectations set by baseline methods in C. albicans and D. melanogaster, it leaves considerable room and need for improvement. Finally, we report that the CAFA community now involves a broad range of participants with expertise in bioinformatics, biological experimentation, biocuration, and bio-ontologies, working together to improve functional annotation, computational function prediction, and our ability to manage big data in the era of large experimental screens.

Residue Cluster Classes: A Unified Protein Representation for Efficient Structural and Functional Classification

“…The latest Critical Assessment for Structure Prediction (CASP) competition for the first time showed that ML (AlphaFold) improved on previous methods using protein sequence features [24], yet the features used by AlphaFold are not the same used by the best models of protein function reported in CAFA contest [8]. Thus, RCC is the first representation of proteins that allows for the efficient modeling of both fundamental aspects of proteins.…”

“…Different approaches have been described to account for such a relationship, including those based on physical and chemical forces [2,3], protein sequence and phylogenies [4][5][6] and others. Machine-learning (ML) based models are currently the best models for predicting 3D protein structures [7] and protein functions [8]. ML models require object representations in the form of a set of features; these features are numeric values; hence, objects are represented by vectors.…”

TANA: efficient approach for predicting protein functions by transferring annotation via alignment networks

“…Indeed, it is, respectively, that the orengo-funfams approach performs well for F max and S min for MF and BP evaluations under the CAFA3 challenge [49]. The approach ranked 2nd and 4th in F max for MF and BP, as well as, respectively, 2nd and 3rd in S min for MF and BP evaluations.…”

“…Using this evaluation, we can conclude that the process of inferring the parent and child terms are over the whole GO. Whereas for the CAFA3 evaluation [49], each single GO term is initially dissociated by their three sub-ontologies. Then, the parent and the child terms are then taken locally into their three respective subontologies.…”

“…Besides, the INGA-Tosatto also performs well for F max and S min for BP and CC predictions. The approach was ranked, respectively, 2nd and 3rd in F max for BP and CC, as well as 2nd in S min for BP and CC evaluations [49].…”