//[[Rcpp::export]]
DataFrame closest_impl(ValrGroupedDataFrame x, ValrGroupedDataFrame y,
IntegerVector grp_idx_x,
IntegerVector grp_idx_y,
const std::string& suffix_x,
const std::string& suffix_y) {
DataFrame df_x = x.data() ;
DataFrame df_y = y.data() ;
// for subsetting / return df
std::vector<int> indices_x ;
std::vector<int> indices_y ;
std::vector<int> overlap_sizes ;
std::vector<int> distance_sizes ;
// set up interval trees for each chromosome and apply closest_grouped
GroupApply(x, y, grp_idx_x, grp_idx_y,
closest_grouped, std::ref(indices_x), std::ref(indices_y),
std::ref(overlap_sizes), std::ref(distance_sizes));
DataFrame subset_x = subset_dataframe(df_x, indices_x) ;
DataFrame subset_y = subset_dataframe(df_y, indices_y) ;
DataFrameBuilder out;
// x names, data
out.add_df(subset_x, suffix_x, false) ;
// y names, data
out.add_df(subset_y, suffix_y, true) ;
// overlaps and distances
out.add_vec(".overlap", wrap(overlap_sizes)) ;
out.add_vec(".dist", wrap(distance_sizes)) ;
auto nrows = subset_x.nrows() ;
auto res = out.format_df(nrows) ;
return res ;
}
//[[Rcpp::export]]
DataFrame intersect_impl(GroupedDataFrame x, GroupedDataFrame y,
const std::string& suffix_x = ".x",
const std::string& suffix_y = ".y") {
// indices for subsetting
std::vector<int> indices_x ;
std::vector<int> indices_y ;
// overlap sizes
std::vector<int> overlap_sizes ;
auto data_x = x.data() ;
auto data_y = y.data() ;
// set up interval trees for each chromosome and apply intersect_group
GroupApply(x, y, intersect_group, std::ref(indices_x), std::ref(indices_y), std::ref(overlap_sizes));
DataFrame subset_x = DataFrameSubsetVisitors(data_x, names(data_x)).subset(indices_x, "data.frame");
DataFrame subset_y = DataFrameSubsetVisitors(data_y, names(data_y)).subset(indices_y, "data.frame");
auto ncol_x = subset_x.size() ;
auto ncol_y = subset_y.size() ;
CharacterVector names(ncol_x + ncol_y) ;
CharacterVector names_x = subset_x.attr("names") ;
CharacterVector names_y = subset_y.attr("names") ;
// replacing y chrom with overlap, same number of cols
List out(ncol_x + ncol_y) ;
// x names, data
for (int i = 0; i < ncol_x; i++) {
auto name_x = as<std::string>(names_x[i]) ;
if (name_x != "chrom") {
name_x += suffix_x ;
}
names[i] = name_x ;
out[i] = subset_x[i] ;
}
// y names, data
for (int i = 0; i < ncol_y; i++) {
auto name_y = as<std::string>(names_y[i]) ;
if (name_y == "chrom") continue ;
name_y += suffix_y ;
names[i + ncol_x - 1] = name_y ;
out[i + ncol_x - 1] = subset_y[i] ;
}
// overlaps
out[ncol_x + ncol_y - 1] = overlap_sizes ;
names[ncol_x + ncol_y - 1] = ".overlap" ;
out.attr("names") = names ;
out.attr("class") = classes_not_grouped() ;
auto nrows = subset_x.nrows() ;
set_rownames(out, nrows) ;
return out ;
}
