improve number formatting

manuscript/_manuscript/_tex/index.tex

@@ -446,13 +446,13 @@ \subsubsection{Validation of expected phenotype-cell type
 \(-log_{10}(\text{p-values})\) into 6 bins. For each HPO-CL branch
 pairing, we then calculated the proportion of on-target cell types
 within each bin. We found that the proportion of on-target cell types
-increased with increasing significance of the association (rho=0.63,
+increased with increasing significance of the association (rho=0.632,
 p=\ensuremath{1.119\times 10^{-6}}). For example, abnormalities of the
-nervous system with \(-log_{10}(\text{p-values}) = 1\), only 16\% of the
-associated cell types were neural cells. Whereas for those with
-\(-log_{10}(\text{p-values}) = 6\), 46\% were neural cells despite the
-fact that this class of cell types only constituted 23\% of the total
-cell types tested (i.e.~the baseline). This shows that the more
+nervous system with \(-log_{10}(\text{p-values}) = 1\), only 15.666\% of
+the associated cell types were neural cells. Whereas for those with
+\(-log_{10}(\text{p-values}) = 6\), 46.104\% were neural cells despite
+the fact that this class of cell types only constituted 23\% of the
+total cell types tested (i.e.~the baseline). This shows that the more
 significant the association, the more likely it is that the cell type is
 on-target.
 
@@ -805,11 +805,11 @@ \subsubsection{Congenital phenotypes are associated with foetal cell
 \end{figure}%
 
 We also found that some branches of the HPO were more commonly enriched
-in foetal cell types compared to others (\(\hat{V}_{Cramer}\)=0.22,
+in foetal cell types compared to others (\(\hat{V}_{Cramer}\)=0.222,
 \(p\)\textless{}\ensuremath{2.225\times 10^{-308}}). See The branch with
 the greatest proportion of fetal cell type enrichments was `Abnormality
-of limbs' (35.46\%), followed by `Growth abnormality' (31.609\%) and
-`Abnormality of the musculoskeletal system' (28.61\%). These results
+of limbs' (35.461\%), followed by `Growth abnormality' (31.609\%) and
+`Abnormality of the musculoskeletal system' (28.611\%). These results
 align well with the fact that physical malformations tend to be
 developmental in origin.
 
@@ -1112,22 +1112,22 @@ \subsection{Discussion}\label{sec-discussion}
 
 Investigating RDs at the level of phenotypes offers several key
 advantages. First, the vast majority of RDs only have one associated
-gene (7671/8631 diseases = 88.877\%). Aggregating gene sets across
-diseases into phenotype-centric ``buckets'' permits sufficiently
-well-powered analyses, with an average of \textasciitilde75.647 genes
-per phenotype (median=7) see Fig.~\ref{fig-diagram}. Second, we
-hypothesise that these phenotype-level gene sets converge on a limited
-number of molecular and cellular pathways. Perturbations to these
-pathways manifest as one or more phenotypes which, when considered
-together, tend to be clinically diagnosed as a certain disease. Third,
-RDs are often highly heterogeneous in their clinical presentation across
-individuals, leading to the creation of an ever increasing number of
-disease subtypes (some of which only have a single documented case). In
-contrast, a phenotype-centric approach enables us to more accurately
-describe a particular individual's version of a disease without relying
-on the generation of additional disease subcategories. By characterising
-an individual's precise phenotypes over time, we may better understand
-the underlying biological mechanisms that have caused their condition.
+gene (7671/8631 diseases = 89\%). Aggregating gene sets across diseases
+into phenotype-centric ``buckets'' permits sufficiently well-powered
+analyses, with an average of \textasciitilde75.647 genes per phenotype
+(median=7) see Fig.~\ref{fig-diagram}. Second, we hypothesise that these
+phenotype-level gene sets converge on a limited number of molecular and
+cellular pathways. Perturbations to these pathways manifest as one or
+more phenotypes which, when considered together, tend to be clinically
+diagnosed as a certain disease. Third, RDs are often highly
+heterogeneous in their clinical presentation across individuals, leading
+to the creation of an ever increasing number of disease subtypes (some
+of which only have a single documented case). In contrast, a
+phenotype-centric approach enables us to more accurately describe a
+particular individual's version of a disease without relying on the
+generation of additional disease subcategories. By characterising an
+individual's precise phenotypes over time, we may better understand the
+underlying biological mechanisms that have caused their condition.
 However, in order to achieve a truly precision-based approach to
 clinical care, we must first characterise the molecular and cellular
 mechanisms that cause the emergence of each phenotype. Here, we provide