Guo et al.

hereby find

truncating mutations not

guilty.

Protein-network-based approaches for identifying candidate genes and prediction of mutation pathogenicity often hinge on the idea that mutations within
genes encoding components of a given pathway should result in the same phenotype.
Goh et al. and Feldman et al. found that protein products of genes associated with similar diseases are more likely to physically interact and form diseasespecic functional modules. On the basis of this commonly accepted guilt-by-association principle methods have not considered the potential differences
in the molecular mechanisms leading to the corresponding disorders for mutations of different inheritance modes and molecular types.
Zhong et al. found that disease mutations could lead to two types of perturbations at the network level: node removal (loss of all known interactions of a
protein) or edgetic perturbation (loss of specic interactions of a protein). They also found that a higher fraction of mutations associated with autosomal-

dominant diseases are in-frame, tend to affect structural proteins, and are likely to affect exposed residues.

Zhong : Solid lines between two nodes represent preserved interactions and dashed lines represent perturbed interactions. Edges are generally
biophysical interactions, but could also be biochemical.
Evaluating over 20,000 cancer-associated mutations to determine whether dominant and recessive mutations are enriched in regions encoding protein
interaction interfaces they found that recessive mutations (not dominant point mutations) are likely to be found in regions encoding interaction interfaces.
Furthermore, dominant truncating mutations are enriched in regions encoding areas between interfaces, suggesting that these mutant genes might encode
functional protein fragments that retain some, but not all, protein interactions.
This finding is particularly jarring as truncating, also referred to as loss-of-function (LoF), mutations are often regarded as knockout mutations in large-scale
mutational screens and genome-sequencing.

However, there are instances reported where mRNAs harboring truncating mutations escape NMD and are translated into proteins with dominantnegative activities. One particularly interesting case study involving SOX10 demonstrated that truncating mutations in different regions of SOX10 confer
distinct neurological phenotypes. Among all SOX10 alleles harboring nonsense or frameshift mutations, transcripts with mutations in exons 3 and 4 are
targeted by NMD, causing a neurological phenotype called Waardenburg-Shah syndrome.
On the other hand, transcripts with mutations in exon 5 escape NMD and lead to a more severe phenotype as a result of the dominantnegative effects
of the translated protein. Furthermore, a recent publication revealed that, contrary to common belief, only a small percentage (16.3%) of LoF alleles
show signicant evidence of NMD . Our results further suggest that it is overly simplistic to consider all truncating mutations as null mutations, given that
a signicant fraction of them do generate functional protein products.
Interestingly, our results show that truncating mutations that lead to functional products are not limited to the extreme C-terminal region of proteins;
many proteins can lose more than two-thirds of their length and still retain specic functions.
Guo et al., Dissecting Disease Inheritance Modes in a Three-Dimensional Protein Network Challenges the Guilt-by-Association Principle . American Journal of Human
Genetics, vol 93, pp. 7889 (2013)

See also: Gillis and Pavlidis, Guilt by Association Is the Exception Rather Than the Rule in Gene Networks. PLOS Computational Biology, vol 8, e1002444 (2012)

Multifunctionality, rather than association, is a primary driver of gene function prediction. Knowledge of the degree of multifunctionality alone can
produce astonishingly strong performance when used as a predictor of gene function.
Multifunctionality is encoded in gene interaction data (such as protein interactions and coexpression networks) and this can feed forward into gene
function prediction algorithms.