Lightnews — Scholar-powered news

Mihaly Badonyi

@mbadonyi.bsky.social

Thanks to #CCG2025 for the opportunity to present our work on `acmgscaler`, a standardised tool to convert functional scores into ACMG/AMP evidence strengths.
#rstats

You can try out the Colab notebook and the R package here: https://github.com/badonyi/acmgscaler

June 13, 2025 at 11:14 AM

Mihaly Badonyi

@mbadonyi.bsky.social

6/ Applying the mLOF score to 2,837 phenotypes across 1,979 Mendelian disease genes, we found that DN and GOF mechanisms account for 48% of phenotypes in dominant genes.

$Caption: mLOF scores reveal the prevalence of molecular mechanisms at the phenotype level. A-B: Distribution of mLOF scores for inheritance- and mechanism-based phenotype classifications (see Table 1 for the description of the abbreviations). N denotes the number of phenotypes in each group. Red line indicates the optimal mLOF score threshold. C: The fractional composition of predicted mechanisms for phenotypes across the indicated categories. The predictions represent the highest-ranking posterior mechanism-specific score for each phenotype. N denotes the number of phenotypes in each group. Bar charts show estimates of genes with both LOF and non-LOF (‘hybrid’) mechanisms. Error bars are 95% credible intervals calculated from a posterior distribution of fractions derived using the bootstrap estimates of the optimal mLOF threshold.$

March 17, 2025 at 12:44 PM

Mihaly Badonyi

@mbadonyi.bsky.social

I also tried with rle(), but couldn't get it right.

$input <- strsplit(readLines('day_12.txt'), '') grid <- do.call(rbind, input) count_sides <- function(seen) { pad <- rbind(FALSE, cbind(FALSE, seen, FALSE), FALSE) count_sides <- function(seen) { sides <- 0 for (i in seq(nrow(seen) - 1)) { row1 <- seen[i, ] row2 <- seen[i + 1, ] sides <- sides + sum(diff(c(FALSE, row1 & !row2)) == 1) + sum(diff(c(FALSE, row2 & !row1)) == 1) } sides } return(count_sides(pad) + count_sides(t(pad))) } block_price <- function(grid, coord, sides = FALSE) { dir <- list(c(-1, 0), c(1, 0), c(0, -1), c(0, 1)) seen <- matrix(FALSE, nrow = nrow(grid), ncol = ncol(grid)) block <- grid[coord[1], coord[2]] coords <- list(coord) area <- perim <- 0 while (length(coords) > 0) { idx <- matrix(coords[[1]], nrow = 1) coords <- coords[-1] if (any(idx <= 0) | any(idx > dim(grid))) { perim <- perim + 1 next } if (grid[idx] != block) { perim <- perim + 1 next } if (seen[idx]) { next } else { seen[idx] <- TRUE } area <- area + 1 coords <- c(coords, lapply(dir, function(x) idx + x)) } if (sides) { perim <- count_sides(seen) } grid[seen] <- 'seen' return(list(grid = grid, price = perim * area)) } calculate_price <- function(grid, sides = FALSE) { price <- 0 while (length(block_coord <- which(grid != 'seen', arr.ind = TRUE)) > 0) { coord <- block_coord[1, , drop = FALSE] block <- block_price(grid, coord, sides) grid <- block$grid price <- price + block$price } return(price) } calculate_price(grid, sides = FALSE) calculate_price(grid, sides = TRUE)$

December 13, 2024 at 2:49 PM

Mihaly Badonyi

@mbadonyi.bsky.social

my day 5 #rstats solution to the 2024 #adventofcode

I have also created an R Colab for these solutions, so R-curious people can run them without having to install R: github.com/badonyi/adve...