Base Excision Repair

For each query set, SNPnexus links genes reported in each dataset to their associated biological pathway(s) (Reactome). First, the variants reported in each dataset are mapped to their corresponding gene. Next, using our MySQL instance of data obtained from Reactome (August 2019), we determine the number of genes within the Reactome universe, the number of genes in each pathway and the identity of genes linked to each pathway. These covariates are used to implement the Fisher's Exact method and determine whether the number of genes affected in each dataset is larger than expected by random chance for each Reactome pathway (Figure 2). Using this method, pathways over-represented within a gene list can be determined by considering both the number of affected genes within the user-defined gene list and the overall number of genes annotated within each Reactome pathway of interest. Analyses of overrepresented signaling pathways can provide useful insights into underlying biological and molecular processes, which can be used to identify genotype-phenotype interactions and inform on mechanisms of disease. Following the determination of statistically-enriched pathways, these results are provided in a tabular format as well as a link to an interactive Voronoi diagram, developed using ReacFoam (12 (link)). These visualizations highlight all pathways that are significantly enriched (P ≤ 0.05) in a yellow color scale. The interactive functions allow users to further examine the results at the individual-pathway level, as well as migrate through pathway hierarchies to determine biological processes affected at the uppermost hierarchical levels (e.g. Base Excision Repair → Base-Excision Repair, AP Site Formation → Depurination; Figure 3).

The software behind the interfaces was programmed in Perl 5.8. Submitted data is read using the CGI module, HTML::Template is used to create the query interface, database connections are made with the DBI module and the DBD::Pg database driver, bar charts are created with the GD module and the Statistics::Basic::Correlation module is used to calculate Pearson correlation for expression data.
For prioritisation, the users can select among different predefined schemes for common approaches, e.g. tissue specific expression or similarity with known disease genes. If a prioritisation approach has been selected, the prioritisation section will open in the interface and the preset weights assigned to each parameter will be filled in by JavaScript. Users are absolutely free to change these settings to values that better reflect their own preconception. After the database was queried, all genes are scored according to their parameters' fulfilment of the settings made in the query interface and the weight assigned to each positive match. The genes are subsequently re-ordered by their scores.
Expression similarity is calculated using Pearson correlation. For this, the mean expression in any available tissue is used. This value can be used for prioritisation (multiplied by the user-defined weight), sorting and filtering. In the latter case, only genes with a correlation higher than the specified factor are shown.
The computation of the similarity of the user specified tissue specific expression is performed by comparison of each tissue's expression/median with the specified value. If the value is above the user input and the operator is ‘greater then’ or if it is below and ‘smaller then’ was selected, a positive score will be generated; in other cases the score will be negative. The score is calculated by division of the real expression/median by the user entered value, if the result is negative, the inverse will be taken. All scores for one gene are added to generate the final similarity score.
Querying fields with a hierarchical structure (e.g. GeneOntology) will also find descendants (subclasses) of an entity, e.g. querying for DNA repair will also find genes, which do not carry this term but its subclasses base-excision repair or mismatch repair instead. To achieve this, a recursive query is carried out using a PL/pgSQL function. Results are written into a temporary table and then used by GeneDistiller to either restrict a query or to highlight values (or their subclasses) matching the user's request.

YiiP mutants were labelled with Alexa fluor 488 fluorescent dye (Invitrogen life Technologies, Carlsbad CA) by adding 2.5 μL of dye from a 16 mM stock solution in DMSO to 200 μL of protein at 1–2 mg/ml in SEC buffer supplemented with 10% glycerol at pH 7. The reactive group of this dye was N-hydroxysuccinimide, which at pH 7 preferentially targets the N-terminus of the polypeptide. Although labeling of lysine residues is also possible, we believe this was minimal due to the low labeling stoichiometries of ~1:1 used for our experiments. Zn²⁺ was removed by adding 0.5 mM EDTA and 0.5 mM TPEN followed by overnight incubation at 4°C. Excess dye and chelated Zn²⁺ were removed using several cycles of dilution with SEC buffer and concentration with a 50 kDa cutoff concentrator (AMICON, Millipore Sigma, Burlington MA). For titration, Zn²⁺ was buffered either by 0.5 mM sodium citrate or 0.2 mM NTA; the total amount of added ZnSO₄ was varied to achieve the desired concentration of free Zn²⁺, according to the program MAXCHELATOR (Bers et al., 1994 (link)). Protein concentration during the titration varied from 8 to 100 nM. After 1:1 mixing of protein and Zn²⁺ solutions, samples were centrifuged for 5 min at 2000xg, then loaded into standard treated capillaries for measurement with a Monolith NT.115 MST instrument (NanoTemper Technologies, South San Francisco, CA); measurements were taken at 37 °C with LED power ranges from 20–60% and medium MST power. Data from three independent titrations were analyzed with the MO.Affinity Analysis software v2.3 using MST on-time of 15 s. For determination of K_d, data was fitted with a curve based on the law of mass action,
$$F\left(C_{Zn}\right)=F_u+\frac{\left(F_b-F_u\right)\ast \left(C_{Zn}+C_P+K_d-\sqrt{{\left(C_{Zn}+C_P+K_d\right)}^2-4\ast C_P\ast C_{Zn}}\right)}{2\ast C_P}$$
where $\text{F}\left({\text{C}}_{\text{Zn}}\right)$ is the fraction bound, ${\text{C}}_{\text{Zn}}$ is concentration of free zinc, ${\text{C}}_{\text{P}}$ is the concentration of YiiP protein, ${\text{F}}_{\text{b}}$ and ${\text{F}}_{\text{u}}$ refer to the normalized fluorescence in the bound and unbound state, and ${\text{K}}_{\text{d}}$ is the affinity constant. Alternatively, the data was fit with the Hill equation to assess cooperativity,
$$F\left(C_{Zn}\right)=F_u=\frac{\left(F_b-F_u\right)}{1+{\left(\frac{E{C}_{50}}{C_{Zn}}\right)}^n}$$
where ${\text{EC}}_{50}$ is the half-maximal effective concentration (akin to ${\text{K}}_{\text{d}}$ ) and $\text{n}$ is the Hill coefficient.