Find open-source science resources

A package to generate high-resolution Venn and Upset plots for genomic interaction data from HiC, ChIA-PET, HiChIP, PLAC-Seq, Hi-TrAC, HiCAR and etc. The package generates plots specifically crafted to eliminate the deceptive visual representation caused by the counts method.

The GenomicInteractionNodes package can import interactions from bedpe file and define the interaction nodes, the genomic interaction sites with multiple interaction loops. The interaction nodes is a binding platform regulates one or multiple genes. The detected interaction nodes will be annotated for downstream validation.

The complexity of high-throughput quantitative omics experiments often leads to low replicates numbers and many missing values. We implemented a new test to simultaneously consider missing values and quantitative changes, which we combined with well-performing statistical tests for high confidence detection of differentially regulated features. The package contains functions to run the test and to visualize the results.

MicrobiotaProcess is an R package for analysis, visualization and biomarker discovery of microbial datasets. It introduces MPSE class, this make it more interoperable with the existing computing ecosystem. Moreover, it introduces a tidy microbiome data structure paradigm and analysis grammar. It provides a wide variety of microbiome data analysis procedures under the unified and common framework (tidy-like framework).

This package is a comprehensive visualization tool specifically designed for exploring phylomorphospace. It not only simplifies the process of generating phylomorphospace, but also enhances it with the capability to add graphic layers to the plot with grammar of graphics to create fully annotated phylomorphospaces. It also provide some utilities to help interpret evolutionary patterns.

dominatR is an R package for quantifying and visualizing feature dominance in datasets. dominatR applies concepts drawn from physics such as center of mass and shannon's entropy to effectively visualize features (e.g. genes) that are present within a specific context or condition. The package integrates, dataframes, matrices and SummerizedExperiment objects and is able to perform common genomic normalization methods. The key aspect is the generation of plots that serve to highlight context-relevant feature dominance.

PIPETS provides statistically robust analysis for 3'-seq/term-seq data. It utilizes a sliding window approach to apply a Poisson Distribution test to identify genomic positions with termination read coverage that is significantly higher than the surrounding signal. PIPETS then condenses proximal signal and produces strand specific results that contain all significant termination peaks.

The arrangement of hypotheses in a hierarchical structure appears in many research fields and often indicates different resolutions at which data can be viewed. This raises the question of which resolution level the signal should best be interpreted on. treeclimbR provides a flexible method to select optimal resolution levels (potentially different levels in different parts of the tree), rather than cutting the tree at an arbitrary level. treeclimbR uses a tuning parameter to generate candidate resolutions and from these selects the optimal one.

countsimQC provides functionality to create a comprehensive report comparing a broad range of characteristics across a collection of count matrices. One important use case is the comparison of one or more synthetic count matrices to a real count matrix, possibly the one underlying the simulations. However, any collection of count matrices can be compared.

Given a sample data table and a design formula, ExploreModelMatrix generates an interactive application for exploration of the resulting design matrix. This can be helpful for interpreting model coefficients and constructing appropriate contrasts in (generalized) linear models. Static visualizations can also be generated.

This package provides extensive functionality for comparing results obtained by different methods for differential expression analysis of RNAseq data. It also contains functions for simulating count data. Finally, it provides convenient interfaces to several packages for performing the differential expression analysis. These can also be used as templates for setting up and running a user-defined differential analysis workflow within the framework of the package.

This package generates microarray quality metrics reports for data in Bioconductor microarray data containers (ExpressionSet, NChannelSet, AffyBatch). One and two color array platforms are supported.

Dot plots of single-cell RNA-seq data allow for an examination of the relationships between cell groupings (e.g. clusters) and marker gene expression. The scDotPlot package offers a unified approach to perform a hierarchical clustering analysis and add annotations to the columns and/or rows of a scRNA-seq dot plot. It works with SingleCellExperiment and Seurat objects as well as data frames.

4way plots enable a comparison of the logFC values from two contrasts of differential gene expression. The gg4way package creates 4way plots using the ggplot2 framework and supports popular Bioconductor objects. The package also provides information about the correlation between contrasts and significant genes of interest.

CATALYST provides tools for preprocessing of and differential discovery in cytometry data such as FACS, CyTOF, and IMC. Preprocessing includes i) normalization using bead standards, ii) single-cell deconvolution, and iii) bead-based compensation. For differential discovery, the package provides a number of convenient functions for data processing (e.g., clustering, dimension reduction), as well as a suite of visualizations for exploratory data analysis and exploration of results from differential abundance (DA) and state (DS) analysis in order to identify differences in composition and expression profiles at the subpopulation-level, respectively.

Takes as input an incomplete perturbation profile and differential gene expression in log odds and infers unobserved perturbations and augments observed ones. The inference is done by iteratively inferring a network from the perturbations and inferring perturbations from the network. The network inference is done by Nested Effects Models.

Mixture Nested Effects Models (mnem) is an extension of Nested Effects Models and allows for the analysis of single cell perturbation data provided by methods like Perturb-Seq (Dixit et al., 2016) or Crop-Seq (Datlinger et al., 2017). In those experiments each of many cells is perturbed by a knock-down of a specific gene, i.e. several cells are perturbed by a knock-down of gene A, several by a knock-down of gene B, ... and so forth. The observed read-out has to be multi-trait and in the case of the Perturb-/Crop-Seq gene are expression profiles for each cell. mnem uses a mixture model to simultaneously cluster the cell population into k clusters and and infer k networks causally linking the perturbed genes for each cluster. The mixture components are inferred via an expectation maximization algorithm.

epiNEM is an extension of the original Nested Effects Models (NEM). EpiNEM is able to take into account double knockouts and infer more complex network signalling pathways. It is tailored towards large scale double knock-out screens.

bnem combines the use of indirect measurements of Nested Effects Models (package mnem) with the Boolean networks of CellNOptR. Perturbation experiments of signalling nodes in cells are analysed for their effect on the global gene expression profile. Those profiles give evidence for the Boolean regulation of down-stream nodes in the network, e.g., whether two parents activate their child independently (OR-gate) or jointly (AND-gate).

The SplicingFactory R package uses transcript-level expression values to analyze splicing diversity based on various statistical measures, like Shannon entropy or the Gini index. These measures can quantify transcript isoform diversity within samples or between conditions. Additionally, the package analyzes the isoform diversity data, looking for significant changes between conditions.

Arc Institute's single-cell foundation model enabling in-context learning at inference time via a novel tabular attention architecture, trained on 150M uniformly-preprocessed cells for generalizing biological effects and generating unseen cell profiles in novel contexts (2025)

Generative AI system for antibiotic discovery that searches billions of synthesizable molecules by combining molecular building blocks through real chemical reactions, experimentally validating novel compounds active against drug-resistant bacteria

Fast spike sorting with drift correction for extracellular electrophysiology, enabling universal neural spike sorting via deep learning on high-density neural probe recordings (MouseLand, 609+ stars)

Provides diagnostics for assessing genomic DNA contamination in RNA-seq data, as well as plots representing these diagnostics. Moreover, the package can be used to get an insight into the strand library protocol used and, in case of strand-specific libraries, the strandedness of the data. Furthermore, it provides functionality to filter out reads of potential gDNA origin.

The Bandle package enables the analysis and visualisation of differential localisation experiments using mass-spectrometry data. Experimental methods supported include dynamic LOPIT-DC, hyperLOPIT, Dynamic Organellar Maps, Dynamic PCP. It provides Bioconductor infrastructure to analyse these data.

The application of 2nd and 3rd generation High Throughput Sequencing (HTS) technologies has deeply reshaped experimental method to investigate microbial communities and obtain a taxonomic and functional profile of the invetigated community. Shotgun Metagenomics allow to quickly obtain a representation of microorganisms genomes characterizing a particular environment. In order to obtain a fast e reliable taxonomic classification of microorganisms genomes we present kMetaShot, an alignment-free taxonomic classifier based on k-mer/minimizer counting.

Tools to normalize (several) Hi-C data from replicates.

Graph objects from pathway topology derived from KEGG, Panther, PathBank, PharmGKB, Reactome SMPDB and WikiPathways databases.

Trackastra is a transformer-based cell-tracking tool for live-cell microscopy. It links already segmented cell instances across time by predicting associations between detections. It supports greedy tracking with or without cell divisions, optional ILP-based linking, pretrained tracking models, and export of tracked masks and lineage information in Cell Tracking Challenge format.

ACME (Algorithms for Calculating Microarray Enrichment) is a set of tools for analysing tiling array ChIP/chip, DNAse hypersensitivity, or other experiments that result in regions of the genome showing "enrichment". It does not rely on a specific array technology (although the array should be a "tiling" array), is very general (can be applied in experiments resulting in regions of enrichment), and is very insensitive to array noise or normalization methods. It is also very fast and can be applied on whole-genome tiling array experiments quite easily with enough memory.

Provides a suite of R functions to analyze gene expression experiments on the OpenArray real-time PCR platform. OAtools fits logistic regressions to fluorescence curves to distinguish between real amplification and false positives. OAtools supports data import, analysis, and visualization through plots and a dynamic HTML report.

This package enables the visualization of functional enrichment results as network graphs. First the package enables the visualization of enrichment results, in a format corresponding to the one generated by gprofiler2, as a customizable Cytoscape network. In those networks, both gene datasets (GO terms/pathways/protein complexes) and genes associated to the datasets are represented as nodes. While the edges connect each gene to its dataset(s). The package also provides the option to create enrichment maps from functional enrichment results. Enrichment maps enable the visualization of enriched terms into a network with edges connecting overlapping genes.

Benchmark evaluating AI agents on 75 curated Kaggle-style ML engineering competitions with reproducible Docker-based grading harness, human baselines, and end-to-end task lifecycle, used as a primary benchmark for autonomous ML research agents (e.g., InternAgent #1 at 36.44%)

The HiCPotts package provides a comprehensive Bayesian framework for analyzing Hi-C interaction data, integrating both spatial and genomic biases within a probabilistic modeling framework. At its core, HiCPotts leverages the Potts model (Wu, 1982)—a well-established graphical model—to capture and quantify spatial dependencies across interaction loci arranged on a genomic lattice. By treating each interaction as a spatially correlated random variable, the Potts model enables robust segmentation of the genomic landscape into meaningful components, such as noise, true signals, and false signals. To model the influence of various genomic biases, HiCPotts employs a regression-based approach incorporating multiple covariates: Genomic distance (D): The distance between interacting loci, recognized as a fundamental driver of contact frequency. GC-content (GC): The local GC composition around the interacting loci, which can influence chromatin structure and interaction patterns. Transposable elements (TEs): The presence and abundance of repetitive elements that may shape contact probability through chromatin organization. Accessibility score (Acc): A measure of chromatin openness, informing how accessible certain genomic regions are to interaction. By embedding these covariates into a hierarchical mixture model, HiCPotts characterizes each interaction’s probability of belonging to one of several latent components. The model parameters, including regression coefficients, zero-inflation parameters (for ZIP/ZINB distributions), and dispersion terms (for NB/ZINB distributions), are inferred via a MCMC sampler. This algorithm draws samples from the joint posterior distribution, allowing for flexible posterior inference on model parameters and hidden states. From these posterior samples, HiCPotts computes posterior means of regression parameters and other quantities of interest. These posterior estimates are then used to calculate the posterior probabilities that assign each interaction to a specific component. The resulting classification sheds light on the underlying structure: distinguishing genuine high-confidence interactions (signal) from background noise and potential false signals, while simultaneously quantifying the impact of genomic biases on observed interaction frequencies. In summary, HiCPotts seamlessly integrates spatial modeling, bias correction, and probabilistic classification into a unified Bayesian inference framework. It provides rich posterior summaries and interpretable, model-based assignments of interaction states, enabling researchers to better understand the interplay between genomic organization, biases, and spatial correlation in Hi-C data.

A user-friendly R package that enables the characterization of each cfDNA fragment overlapping one or multiple mutations of interest, starting from a sequencing file containing aligned reads (BAM file). fRagmentomics supports multiple mutation input formats (e.g., VCF, TSV, or string "chr:pos:ref:alt" representation), accommodates one-based and zero-based genomic conventions, handles mutation representation ambiguities, and accepts any reference file and species in FASTA format. For each cfDNA fragment, fRagmentomics outputs its size, its 3' and 5' sequences, and its mutational status. Optionally, when users set apply_bcftools_norm = TRUE, fRagmentomics invokes the external command-line tool bcftools norm to left-align and normalize variants. If bcftools is not found on the system PATH while this option is enabled, the function errors. The package does not install external software; see the INSTALL file for per-OS instructions.

toscca is an R package to perform Thresholded Ordered Sparse Canonical Correlation Analysis (TOSCCA).

Provides delayed computation of a matrix of residuals after fitting a linear model to each column of an input matrix. Also supports partial computation of residuals where selected factors are to be preserved in the output matrix. Implements a number of efficient methods for operating on the delayed matrix of residuals, most notably matrix multiplication and calculation of row/column sums or means.

A tool that enables in silico identification, integration, and modeling of mRNA features that influence post-transcriptional regulation of gene expression at a transcriptome-wide scale.

KnowYourCG (KYCG) is a supervised learning framework designed for the functional analysis of DNA methylation data. Unlike existing tools that focus on genes or genomic intervals, KnowYourCG directly targets CpG dinucleotides, featuring automated supervised screenings of diverse biological and technical influences, including sequence motifs, transcription factor binding, histone modifications, replication timing, cell-type-specific methylation, and trait-epigenome associations. KnowYourCG addresses the challenges of data sparsity in various methylation datasets, including low-pass Nanopore sequencing, single-cell DNA methylomes, 5-hydroxymethylation profiles, spatial DNA methylation maps, and array-based datasets for epigenome-wide association studies and epigenetic clocks (<doi:10.1126/sciadv.adw3027>). KnowYourCG v2, a command-line implementation in C, is available at <https://github.com/zhou-lab/kycg>.

ClonalSim generates realistic mutational profiles of tumor samples with hierarchical clonal structure. It simulates founder, shared, and private mutations with biologically realistic noise models including intra-tumor heterogeneity (Beta distribution) and technical sequencing noise (negative binomial depth variation, binomial read sampling, base errors). The package is designed for benchmarking variant callers, testing clonal deconvolution algorithms, and teaching tumor heterogeneity concepts.

DuplexDiscovereR is a package designed for analyzing data from RNA cross-linking and proximity ligation protocols such as SPLASH, PARIS, LIGR-seq, and others. DuplexDiscovereR accepts input in the form of chimerically or split-aligned reads. It includes procedures for alignment classification, filtering, and efficient clustering of individual chimeric reads into duplex groups (DGs). Once DGs are identified, the package predicts RNA duplex formation and their hybridization energies. Additional metrics, such as p-values for random ligation hypothesis or mean DG alignment scores, can be calculated to rank final set of RNA duplexes. Data from multiple experiments or replicates can be processed separately and further compared to check the reproducibility of the experimental method.

Assigning probability scores to protein interactions captured in affinity purification mass spectrometry (AP-MS) expriments to infer protein-protein interactions. The output would facilitate non-specific background removal as contaminants are commonly found in AP-MS data.

Hilbert curve is a type of space-filling curves that fold one dimensional axis into a two dimensional space, but with still preserves the locality. This package aims to provide an easy and flexible way to visualize data through Hilbert curve.

hammers is a utilities suite for scRNA-seq data analysis compatible with both Seurat and SingleCellExperiment. It provides simple tools to address tasks such as retrieving aggregate gene statistics, finding and removing rare genes, performing representation analysis, computing the center of mass for the expression of a gene of interest in low-dimensional space, and calculating silhouette and cluster-normalized silhouette.

Protein structure prediction

NormalyzerDE provides screening of normalization methods for LC-MS based expression data. It calculates a range of normalized matrices using both existing approaches and a novel time-segmented approach, calculates performance measures and generates an evaluation report. Furthermore, it provides an easy utility for Limma- or ANOVA- based differential expression analysis.

The package pRolocGUI comprises functions to interactively visualise spatial proteomics data on the basis of pRoloc, pRolocdata and shiny.

The pRoloc package implements machine learning and visualisation methods for the analysis and interogation of quantitiative mass spectrometry data to reliably infer protein sub-cellular localisation.

Distributional flow matching model for robust single-cell perturbation prediction, modeling the full distribution of perturbed cellular expression profiles conditioned on control states via PAD-Transformer and multi-kernel MMD regularization; reduces MSE by 19.6% over the strongest baseline in combinatorial settings (Westlake University, 41+ stars, MIT License)

FutureHouse's end-to-end scientific discovery multi-agent system orchestrating literature search (Crow/Falcon) and data analysis (Finch) agents, first AI-generated drug discovery identifying ripasudil as novel dry AMD therapeutic (2025)