The problems of such a methodology are the finite number of domains available for annotating a protein sequence and the limitation in the choice of the initial protein sequences that are used to define a domain family. They give a broad view of the evolutionary mechanisms at play in the emergence, disappearance and re-use of the evolutionary units that are protein domains. Several protein domain databases 5, 6, 7, 8, 15 have been developed allowing automatic protein annotation. In this sequence-based definition, the foldability is an optional characteristic however, there is a good general agreement between the two types of domain definitions, as assessed by the comparison of sequence-based and structure-based domain databases 13, 14. In the present study, references to protein domain will be based on this last, sequence-based definition, as it allows the largest coverage of the protein sequence space. If the protein sequence information is used, a protein domain then corresponds to the amino acid sequence part that is evolutionary conserved and for which similar sequences can be found in other proteins of the same or in different organisms 8. If structural data are considered, a protein domain can be defined as a basic unit that has the ability to fold independently and is characterized by an overall compactness, driven by the presence of a hydrophobic core, or by that of metal ions or disulfide bridges 11, 12. Two distinct definitions of a protein domain co-exist regarding the type of molecular data used for their identification. those basic units from which proteins are made, are the essential blocks of protein evolution 4, 9, 10. It is well established that domains, i.e.

The task is challenging but the community effort to give access to all protein sequences led to the fast emergence of several resources for protein annotation that are still intensely maintained up-to-date 5, 6, 7, 8. However, the development of powerful bioinformatics methods gave the opportunity to automatically classify and annotate protein sequences. The protein universe cannot be explored exhaustively in an expert-based approach, due to the huge amount of data available today. Moreover, deciphering the rules behind the apparition of new proteins and their conservation in an organism may lead to develop improved methods for protein design. Understanding how the current protein universe appeared and evolved are central questions in evolutionary molecular biology 2, 3, 4.

The protein universe is a concept corresponding to the set of all possible protein sequences that can be found in all living organisms 1. Based on these observations, we emphasize that domain annotation methodologies can still be improved to fully apprehend and decipher the molecular evolution of the protein universe. These features make them hard to relate to known families. The un-annotated domains of the dark proteome universe display specific features relative to database domains: shorter length, non-canonical content and particular topology in hydrophobic residues, higher propensity for disorder, and a higher energy. The dark foldable domains were analyzed to understand what make them different from domains stored in databases and thus difficult to annotate. These un-annotated foldable domains were grouped using a combination of remote homology searches and domain annotations, leading to define different levels of darkness. In this study, we quantify the amount of foldable domains within the dark proteome by using the hydrophobic cluster analysis methodology. The un-annotated part of the protein universe is referred to as the dark proteome and remains poorly characterized. However, some part of sequences and many entire sequences remain un-annotated despite a converging number of domain families. A way to explore it is by taking into account the domain content of the proteins. The protein universe corresponds to the set of all proteins found in all organisms.

Large hydrophobic amino acids