Find open-source science resources

Implements a DelayedArray backend for reading and writing dense or sparse arrays in the TileDB format. The resulting TileDBArrays are compatible with all Bioconductor pipelines that can accept DelayedArray instances.

Offline-first scientific writing workspace powered by Claude, integrating LaTeX, Python, and 100+ scientific skills with local execution, Zotero integration, and privacy-focused design (2026)

Improved equivariant Transformer for 3D atomic graphs (ICLR2024)

Provides a consistent C++ class interface for reading from a variety of commonly used matrix types. Ordinary matrices and several sparse/dense Matrix classes are directly supported, along with a subset of the delayed operations implemented in the DelayedArray package. All other matrix-like objects are supported by calling back into R.

Parameter/topology editor and molecular simulator with visualization capability.

Medical time series foundation model pretrained on 454B time points from heterogeneous clinical corpora spanning ICU physiological signals and hospital EHR, with continuous-time rotary positional encoding, frequency-specialized Mixture-of-Experts, and neural ODE extrapolation for zero-shot forecasting across irregular and multimodal temporal health data (Microsoft, 399+ stars, MIT License)

Lightweight Expression displaYer (plotter / viewer) of SummarizedExperiment object in R. This package provides a quick and easy Shiny-based GUI to empower a user to use a SummarizedExperiment object to view

Point mutations occurring in a genome can be divided into 96 categories based on the base being mutated, the base it is mutated into and its two flanking bases. Therefore, for any patient, it is possible to represent all the point mutations occurring in that patient's tumor as a vector of length 96, where each element represents the count of mutations for a given category in the patient. A mutational signature represents the pattern of mutations produced by a mutagen or mutagenic process inside the cell. Each signature can also be represented by a vector of length 96, where each element represents the probability that this particular mutagenic process generates a mutation of the 96 above mentioned categories. In this R package, we provide a set of functions to extract and visualize the mutational signatures that best explain the mutation counts of a large number of patients.

Mutations that rapidly accumulate in viral genomes during a pandemic can be used to track the evolution of the virus and, accordingly, unravel the viral infection network. To this extent, sequencing samples of the virus can be employed to estimate models from genomic epidemiology and may serve, for instance, to estimate the proportion of undetected infected people by uncovering cryptic transmissions, as well as to predict likely trends in the number of infected, hospitalized, dead and recovered people. VERSO is an algorithmic framework that processes variants profiles from viral samples to produce phylogenetic models of viral evolution. The approach solves a Boolean Matrix Factorization problem with phylogenetic constraints, by maximizing a log-likelihood function. VERSO includes two separate and subsequent steps; in this package we provide an R implementation of VERSO STEP 1.

Cancer is a genetic disease caused by somatic mutations in genes controlling key biological functions such as cellular growth and division. Such mutations may arise both through cell-intrinsic and exogenous processes, generating characteristic mutational patterns over the genome named mutational signatures. The study of mutational signatures have become a standard component of modern genomics studies, since it can reveal which (environmental and endogenous) mutagenic processes are active in a tumor, and may highlight markers for therapeutic response. Mutational signatures computational analysis presents many pitfalls. First, the task of determining the number of signatures is very complex and depends on heuristics. Second, several signatures have no clear etiology, casting doubt on them being computational artifacts rather than due to mutagenic processes. Last, approaches for signatures assignment are greatly influenced by the set of signatures used for the analysis. To overcome these limitations, we developed RESOLVE (Robust EStimation Of mutationaL signatures Via rEgularization), a framework that allows the efficient extraction and assignment of mutational signatures. RESOLVE implements a novel algorithm that enables (i) the efficient extraction, (ii) exposure estimation, and (iii) confidence assessment during the computational inference of mutational signatures.

Single-cell RNA-seq technologies enable high throughput gene expression measurement of individual cells, and allow the discovery of heterogeneity within cell populations. Measurement of cell-to-cell gene expression similarity is critical for the identification, visualization and analysis of cell populations. However, single-cell data introduce challenges to conventional measures of gene expression similarity because of the high level of noise, outliers and dropouts. We develop a novel similarity-learning framework, SIMLR (Single-cell Interpretation via Multi-kernel LeaRning), which learns an appropriate distance metric from the data for dimension reduction, clustering and visualization.

LACE is an algorithmic framework that processes single-cell somatic mutation profiles from cancer samples collected at different time points and in distinct experimental settings, to produce longitudinal models of cancer evolution. The approach solves a Boolean Matrix Factorization problem with phylogenetic constraints, by maximizing a weighed likelihood function computed on multiple time points.

DeepConsensus uses gap-aware sequence transformers to correct errors in Pacific Biosciences (PacBio) Circular Consensus Sequencing (CCS) data.

Tools to harmonize bulk RNA-seq matrices, optionally apply batch correction, and train cross-validated classification models using ranger, glmnet, or xgboost. Supports leakage-safe feature selection, permutation importance, SHAP-based interpretability, and calibration methods (Platt or isotonic). Provides stability metrics across folds, embeddings (PCA/UMAP), ROC visualization, SHAP dependence plots, and tidy ranked-gene tables for downstream analysis.

Accessible protein design platform via Google Colab integrating AlphaFold2, RoseTTAFold, and ProteinMPNN for de novo hallucination, fixed backbone design, and binder design (Sergey Ovchinnikov, 2022+)

Agent skill for AI-assisted scientific manuscript writing review distilled from Stanford's *Writing in the Sciences* course, performing five sequential editorial audit passes on clarity, voice, structure, consistency, and integrity (2026)

Provides a consistent interface for downloading, storing, and accessing immune receptor (TCR/BCR) and HLA sequences from IMGT, IPD-IMGT/HLA, and OGRDB (AIRR-C). Supports export to popular analysis tools including MiXCR, TRUST4, Cell Ranger, and IgBLAST. This package serves as a core dependency for immunogenomics packages, ensuring reliable and high-quality sequence access with local caching for reproducibility.

Implementation of the Ibex algorithm for single-cell embedding based on BCR sequences. The package includes a standalone function to encode BCR sequence information by amino acid properties or sequence order using tensorflow-based autoencoder. In addition, the package interacts with SingleCellExperiment or Seurat data objects.

HiCool provides an R interface to process and normalize Hi-C paired-end fastq reads into .(m)cool files. .(m)cool is a compact, indexed HDF5 file format specifically tailored for efficiently storing HiC-based data. On top of processing fastq reads, HiCool provides a convenient reporting function to generate shareable reports summarizing Hi-C experiments and including quality controls.

Provides R wrappers of several on-target and off-target scoring methods for CRISPR guide RNAs (gRNAs). The following nucleases are supported: SpCas9, AsCas12a, enAsCas12a, and RfxCas13d (CasRx). The available on-target cutting efficiency scoring methods are RuleSet1, RuleSet3, DeepHF, enPAM+GB, and CRISPRscan. Both the CFD and MIT scoring methods are available for off-target specificity prediction. The package also provides a Lindel-derived score to predict the probability of a gRNA to produce indels inducing a frameshift for the Cas9 nuclease. Note that DeepHF and enPAM+GB are not available on Windows machines.

Provides a user-friendly interface to map on-targets and off-targets of CRISPR gRNA spacer sequences using bwa. The alignment is fast, and can be performed using either commonly-used or custom CRISPR nucleases. The alignment can work with any reference or custom genomes. Currently not supported on Windows machines.

The package provides S4 classes and methods to filter, summarise and visualise genetic variation data stored in VCF files. In particular, the package extends the FilterRules class (S4Vectors package) to define news classes of filter rules applicable to the various slots of VCF objects. Functionalities are integrated and demonstrated in a Shiny web-application, the Shiny Variant Explorer (tSVE).

Complex heatmaps are efficient to visualize associations between different sources of data sets and reveal potential patterns. Here the ComplexHeatmap package provides a highly flexible way to arrange multiple heatmaps and supports various annotation graphics.

LLM agent framework for Earth Observation with 104 specialized tools across 5 functional kits

Account for missing values in label-free mass spectrometry data without imputation. The package implements a probabilistic dropout model that ensures that the information from observed and missing values are properly combined. It adds empirical Bayesian priors to increase power to detect differentially abundant proteins.

RegulonDB has collected, harmonized and centralized data from hundreds of experiments for nearly two decades and is considered a point of reference for transcriptional regulation in Escherichia coli K12. Here, we present the regutools R package to facilitate programmatic access to RegulonDB data in computational biology. regutools provides researchers with the possibility of writing reproducible workflows with automated queries to RegulonDB. The regutools package serves as a bridge between RegulonDB data and the Bioconductor ecosystem by reusing the data structures and statistical methods powered by other Bioconductor packages. We demonstrate the integration of regutools with Bioconductor by analyzing transcription factor DNA binding sites and transcriptional regulatory networks from RegulonDB. We anticipate that regutools will serve as a useful building block in our progress to further our understanding of gene regulatory networks.

The iModMix network-based method offers an integrated framework for analyzing multi-omics data, including metabolomics, proteomics, and transcriptomics data, enabling the exploration of intricate molecular associations within heterogeneous biological systems.

This package provides helper functions for working with multiple Visium capture areas that overlap each other. This package was developed along with the companion example use case data available from https://github.com/LieberInstitute/visiumStitched_brain. visiumStitched prepares SpaceRanger (10x Genomics) output files so you can stitch the images from groups of capture areas together with Fiji. Then visiumStitched builds a SpatialExperiment object with the stitched data and makes an artificial hexagonal grid enabling the seamless use of spatial clustering methods that rely on such grid to identify neighboring spots, such as PRECAST and BayesSpace. The SpatialExperiment objects created by visiumStitched are compatible with spatialLIBD, which can be used to build interactive websites for stitched SpatialExperiment objects. visiumStitched also enables casting SpatialExperiment objects as Seurat objects.

Tool to build force field input files for molecular simulation.

The recount3 package enables access to a large amount of uniformly processed RNA-seq data from human and mouse. You can download RangedSummarizedExperiment objects at the gene, exon or exon-exon junctions level with sample metadata and QC statistics. In addition we provide access to sample coverage BigWig files.

scToppR provides an easy-to-use API wrapper for the ToppGene web platform, used for gene ontology and functional enrichment research. The package also integrates visualization tools, making it a convenient tool directly connecting ToppGene to code-based workflows in R. The tool can also easily save results into different formats.

mzR provides a unified API to the common file formats and parsers available for mass spectrometry data. It comes with a subset of the proteowizard library for mzXML, mzML and mzIdentML. The netCDF reading code has previously been used in XCMS.

Explore and download data from the recount project available at https://jhubiostatistics.shinyapps.io/recount/. Using the recount package you can download RangedSummarizedExperiment objects at the gene, exon or exon-exon junctions level, the raw counts, the phenotype metadata used, the urls to the sample coverage bigWig files or the mean coverage bigWig file for a particular study. The RangedSummarizedExperiment objects can be used by different packages for performing differential expression analysis. Using http://bioconductor.org/packages/derfinder you can perform annotation-agnostic differential expression analyses with the data from the recount project as described at http://www.nature.com/nbt/journal/v35/n4/full/nbt.3838.html.

RNA abundance and cell size parameters could improve RNA-seq deconvolution algorithms to more accurately estimate cell type proportions given the different cell type transcription activity levels. A Total RNA Expression Gene (TREG) can facilitate estimating total RNA content using single molecule fluorescent in situ hybridization (smFISH). We developed a data-driven approach using a measure of expression invariance to find candidate TREGs in postmortem human brain single nucleus RNA-seq. This R package implements the method for identifying candidate TREGs from snRNA-seq data.

Generate HTML or PDF reports to explore a set of regions such as the results from annotation-agnostic expression analysis of RNA-seq data at base-pair resolution performed by derfinder. You can also create reports for DESeq2 or edgeR results.

The qsvaR package contains functions for removing the effect of degration in rna-seq data from postmortem brain tissue. The package is equipped to help users generate principal components associated with degradation. The components can be used in differential expression analysis to remove the effects of degradation.

This package provides plotting functions for results from the derfinder package. This helps separate the graphical dependencies required for making these plots from the core functionality of derfinder.

This package provides functions for annotation-agnostic differential expression analysis of RNA-seq data. Two implementations of the DER Finder approach are included in this package: (1) single base-level F-statistics and (2) DER identification at the expressed regions-level. The DER Finder approach can also be used to identify differentially bounded ChIP-seq peaks.

Helper package for speeding up the derfinder package when using multiple cores. This package is particularly useful when using BiocParallel and it helps reduce the time spent loading the full derfinder package when running the F-statistics calculation in parallel.

Builds prediction interval for cell type annotation using conformal inference and conformal risk control. It provides two main methods. The first one gives prediction intervals with coverage guarantees based on standard conformal inference. The second one instead gives hierarchical prediction intervals that are consistent with the cell ontology.

GFF and GTF file manipulation and interconversion.

Differential open reading frame (ORF) translation analysis framework for ribosome profiling (Ribo-seq) with matched RNA-seq. Implements (i) Differential ORF Usage (DOU), a beta-binomial generalized linear model that models the expected proportion of Ribo-seq versus RNA-seq reads mapping to each ORF within a gene, and (ii) ORF-level Differential Translation Efficiency (DTE), a negative binomial GLM that capture changes in translation efficiency of individual ORFs across experimental conditions. Supports ORF-level read summarization for bulk and single-cell Ribo-seq.

Baidu's open-source reproduction of AlphaFold3 in PaddlePaddle, providing pretrained weights and inference pipelines for unified biomolecular structure prediction across proteins, nucleic acids, ligands, ions, and post-translational modifications within the PaddleHelix biocomputing platform (Baidu, bioRxiv 2024)

cytofQC is a package for initial cleaning of CyTOF data. It uses a semi-supervised approach for labeling cells with their most likely data type (bead, doublet, debris, dead) and the probability that they belong to each label type. This package does not remove data from the dataset, but provides labels and information to aid the data user in cleaning their data. Our algorithm is able to distinguish between doublets and large cells.

Google DeepMind's diffusion-based ensemble weather forecasting model at 0.25° resolution, outperforming ECMWF ENS on 97.2% of targets up to 15 days ahead, with open-source code and weights (Nature 2024)

Inference of ligand-receptor (LR) interactions from bulk expression (transcriptomics/proteomics) data, or spatial transcriptomics. BulkSignalR bases its inferences on the LRdb database included in our other package, SingleCellSignalR available from Bioconductor. It relies on a statistical model that is specific to bulk data sets. Different visualization and data summary functions are proposed to help navigating prediction results.

Google Colab-based no-code toolbox democratizing deep learning in microscopy for biologists without programming experience, enabling AI-powered image segmentation, denoising, super-resolution, and object tracking across diverse imaging modalities (Henriques Lab, 640+ stars)

First architecture deeply integrating a DNA foundation model with an LLM for multimodal biological reasoning, achieving 98% accuracy on KEGG disease pathway prediction and 15%+ average gains on variant effect prediction with interpretable step-by-step reasoning traces (bowang-lab, 390+ stars)

A differential abundance analysis for the comparison of two or more conditions. Useful for analyzing data from standard RNA-seq or meta-RNA-seq assays as well as selected and unselected values from in-vitro sequence selections. Uses a Dirichlet-multinomial model to infer abundance from counts, optimized for three or more experimental replicates. The method infers biological and sampling variation to calculate the expected false discovery rate, given the variation, based on a Wilcoxon Rank Sum test and Welch's t-test (via aldex.ttest), a Kruskal-Wallis test (via aldex.kw), a generalized linear model (via aldex.glm), or a correlation test (via aldex.corr). All tests report predicted p-values and posterior Benjamini-Hochberg corrected p-values. ALDEx2 also calculates expected standardized effect sizes for paired or unpaired study designs. ALDEx2 can now be used to estimate the effect of scale on the results and report on the scale-dependent robustness of results.

Allen Institute for AI's global geospatial foundation model for satellite imagery analysis, enabling large-scale mapping of buildings, wind turbines, trees, and land cover from Sentinel-2 data with open-source weights and inference tools (2024)