Find open-source science resources

Neural optical understanding for academic documents, transforms scientific PDFs to Markdown with mathematical formula support

Generate comprehensive reviews from arXiv papers and convert to blog posts

Microsoft's AI-powered ab initio biomolecular dynamics simulation achieving quantum-mechanical accuracy for proteins with 10,000+ atoms, orders of magnitude faster than DFT using protein fragmentation and ML force fields (Nature 2024)

A database system designed to store, organize, and manage large-scale nucleotide sequencing read data (like PacBio reads) for the Dazzler genome assembler

This package defines a custom landing page for an iSEE app interfacing with the Bioconductor ExperimentHub. The landing page allows users to browse the ExperimentHub, select a data set, download and cache it, and import it directly into a Bioconductor iSEE app.

Versatile multi-temporal geospatial foundation model for Earth observation, built on a ViT-based masked autoencoder with 3D spatiotemporal patch embeddings and geolocation/temporal metadata encoding; pretrained on 4.2M global time-series samples from NASA's Harmonized Landsat and Sentinel-2 archive at 30m resolution, with 300M/600M parameter variants and fine-tuning configs for flood detection, wildfire scar, landslide detection, crop segmentation, land cover, and biomass estimation (258+ stars, MIT License)

Equivariant graph attention Transformer (ICLR2023)

The eiR package provides utilities for accelerated structure similarity searching of very large small molecule data sets using an embedding and indexing approach.

A toolbox for sparse contrastive principal component analysis (scPCA) of high-dimensional biological data. scPCA combines the stability and interpretability of sparse PCA with contrastive PCA's ability to disentangle biological signal from unwanted variation through the use of control data. Also implements and extends cPCA.

Extension of ProteinMPNN for protein sequence design in the context of small-molecule ligands, metal ions, and nucleic acids, enabling binding site engineering and co-factor redesign (Baker Lab)

A system for rapidly aligning entire genomes, whether in complete or draft form.

Physics-AI hybrid modeling for fine-grained weather forecasting (NeurIPS'24)

An easy to use tool that can compare splicing events in tumor and normal tissue samples using either a user generated matrix, or data from The Cancer Genome Atlas (TCGA). This package generates a matrix of splicing outliers that are significantly over or underexpressed in tumors samples compared to normal denoted by chromosome location. The package also will calculate the splicing burden in each tumor and characterize the types of splicing events that occur.

Comprehensive survey of foundation models for weather and climate data understanding

This package contains R functions to predict biological variables to from placnetal DNA methylation data generated from infinium arrays. This includes inferring ethnicity/ancestry, gestational age, and cell composition from placental DNA methylation array (450k/850k) data.

Geometric deep learning model predicting transcriptional outcomes of novel single- and multi-gene perturbations using gene–gene knowledge graphs, 40% higher precision than prior methods on combinatorial perturbation prediction (Stanford, Nature Biotechnology 2024)

clevRvis provides a set of visualization techniques for clonal evolution. These include shark plots, dolphin plots and plaice plots. Algorithms for time point interpolation as well as therapy effect estimation are provided. Phylogeny-aware color coding is implemented. A shiny-app for generating plots interactively is additionally provided.

FeatSeekR performs unsupervised feature selection using replicated measurements. It iteratively selects features with the highest reproducibility across replicates, after projecting out those dimensions from the data that are spanned by the previously selected features. The selected a set of features has a high replicate reproducibility and a high degree of uniqueness.

LLM for scientific research papers

Open-source medical large language model for complex clinical reasoning, extending the o1 long-chain-of-thought paradigm to biomedical question answering and diagnostic inference (FreedomIntelligence, 1.3K+ stars)

Kolmogorov-Arnold Networks with learnable activation functions on edges instead of fixed node activations, achieving strong performance in function fitting, PDE solving, and scientific discovery with enhanced interpretability as an alternative to MLPs (MIT, 16.3K+ stars, 2024)

Provides delayed computation of a matrix of scaled and centered values. The result is equivalent to using the scale() function but avoids explicit realization of a dense matrix during block processing. This permits greater efficiency in common operations, most notably matrix multiplication.

Polytect is an advanced computational tool designed for the analysis of multi-color digital PCR data. It provides automatic clustering and labeling of partitions into distinct groups based on clusters first identified by the flowPeaks algorithm. Polytect is particularly useful for researchers in molecular biology and bioinformatics, enabling them to gain deeper insights into their experimental results through precise partition classification and data visualization.

Git repo for Bio::DB::HTS module on CPAN, providing Perl links into HTSlib

A module for solving and visualizing the Schrödinger equation.

Comprehensive toolkit for high-quality PDF content extraction with layout detection, formula recognition, and OCR

Implements a DelayedArray of random values where the realization of the sampled values is delayed until they are needed. Reproducible sampling within any subarray is achieved by chunking where each chunk is initialized with a different random seed and stream. The usual distributions in the stats package are supported, along with scalar, vector and arrays for the parameters.

metagenomeSeq is designed to determine features (be it Operational Taxanomic Unit (OTU), species, etc.) that are differentially abundant between two or more groups of multiple samples. metagenomeSeq is designed to address the effects of both normalization and under-sampling of microbial communities on disease association detection and the testing of feature correlations.

Single-cell transformer foundation model pretrained on 104M human transcriptomes via masked gene prediction, enabling transfer learning for cell type classification, gene network analysis, and in silico perturbation with limited labeled data (Nature 2023, V2 2024)

LLM agents across scientific domains

Rank results by confident effect sizes, while maintaining False Discovery Rate and False Coverage-statement Rate control. Topconfects is an alternative presentation of TREAT results with improved usability, eliminating p-values and instead providing confidence bounds. The main application is differential gene expression analysis, providing genes ranked in order of confident log2 fold change, but it can be applied to any collection of effect sizes with associated standard errors.

PAST takes GWAS output and assigns SNPs to genes, uses those genes to find pathways associated with the genes, and plots pathways based on significance. Implements methods for reading GWAS input data, finding genes associated with SNPs, calculating enrichment score and significance of pathways, and plotting pathways.

A set of tools for interacting with the Food-Biomarker Ontology (FOBI). A collection of basic manipulation tools for biological significance analysis, graphs, and text mining strategies for annotating nutritional data.

NetPathMiner is a general framework for network path mining using genome-scale networks. It constructs networks from KGML, SBML and BioPAX files, providing three network representations, metabolic, reaction and gene representations. NetPathMiner finds active paths and applies machine learning methods to summarize found paths for easy interpretation. It also provides static and interactive visualizations of networks and paths to aid manual investigation.

Define a relatively light class for managing Xenium data using Bioconductor. Address use of parquet for coordinates, SpatialExperiment for assay and sample data. Address serialization and use of cloud storage.

This package is for designing Crispr/Cas9 and Prime Editing experiments. It contains functions to (1) define and transform genomic targets, (2) find spacers (4) count offtarget (mis)matches, and (5) compute Doench2016/2014 targeting efficiency. Care has been taken for multicrispr to scale well towards large target sets, enabling the design of large Crispr/Cas9 libraries.

The POMA package offers a comprehensive toolkit designed for omics data analysis, streamlining the process from initial visualization to final statistical analysis. Its primary goal is to simplify and unify the various steps involved in omics data processing, making it more accessible and manageable within a single, intuitive R package. Emphasizing on reproducibility and user-friendliness, POMA leverages the standardized SummarizedExperiment class from Bioconductor, ensuring seamless integration and compatibility with a wide array of Bioconductor tools. This approach guarantees maximum flexibility and replicability, making POMA an essential asset for researchers handling omics datasets. See https://github.com/pcastellanoescuder/POMAShiny. Paper: Castellano-Escuder et al. (2021) <doi:10.1371/journal.pcbi.1009148> for more details.

stJoincount facilitates the application of join count analysis to spatial transcriptomic data generated from the 10x Genomics Visium platform. This tool first converts a labeled spatial tissue map into a raster object, in which each spatial feature is represented by a pixel coded by label assignment. This process includes automatic calculation of optimal raster resolution and extent for the sample. A neighbors list is then created from the rasterized sample, in which adjacent and diagonal neighbors for each pixel are identified. After adding binary spatial weights to the neighbors list, a multi-categorical join count analysis is performed to tabulate "joins" between all possible combinations of label pairs. The function returns the observed join counts, the expected count under conditions of spatial randomness, and the variance calculated under non-free sampling. The z-score is then calculated as the difference between observed and expected counts, divided by the square root of the variance.

Feature rankings can be distorted by a single case in the context of high-dimensional data. The cases exerts abnormal influence on feature rankings are called influential points (IPs). The package aims at detecting IPs based on case deletion and quantifies their effects by measuring the rank changes (DOI:10.48550/arXiv.2303.10516). The package applies a novel rank comparing measure using the adaptive weights that stress the top-ranked important features and adjust the weights to ranking properties.

Cis-mQTL mapping protocol for placental methylome using TensorQTL, providing a step-by-step guide for genotype QC, imputation, and analysis with detailed scripts and commands.

SPsimSeq uses a specially designed exponential family for density estimation to constructs the distribution of gene expression levels from a given real RNA sequencing data (single-cell or bulk), and subsequently simulates a new dataset from the estimated marginal distributions using Gaussian-copulas to retain the dependence between genes. It allows simulation of multiple groups and batches with any required sample size and library size.

This package detects significant differentially methylated regions (for both qualitative and quantitative traits), using a scan statistic with underlying Poisson heuristics. The scan statistic will depend on a sequence of window sizes (# of CpGs within each window) and on a threshold for each window size. This threshold can be calculated by three different means: i) analytically using Siegmund et.al (2012) solution (preferred), ii) an important sampling as suggested by Zhang (2008), and a iii) full MCMC modeling of the data, choosing between a number of different options for modeling the dependency between each CpG.

This package contains diverse functionality to extend the usage of the iSEE package, including additional classes for the panels or modes facilitating the analysis of differential expression results. This package does not perform differential expression. Instead, it provides methods to embed precomputed differential expression results in a SummarizedExperiment object, in a manner that is compatible with interactive visualisation in iSEE applications.

This package integrates colocalization probabilities from colocalization analysis with transcriptome-wide association study (TWAS) scan summary statistics to implicate genes that may be biologically relevant to a complex trait. The probabilistic framework implemented in this package constrains the TWAS scan z-score-based likelihood using a gene-level colocalization probability. Given gene set annotations, this package can estimate gene set enrichment using posterior probabilities from the TWAS-colocalization integration step.

Large-scale biomolecular instruction dataset for chemistry/biology LLMs (ICLR2024)

Large Language Models for automated open-domain scientific hypotheses discovery (ACL 2024, ICML Best Poster)

Chemical language model

Provides a user-friendly interface to map on-targets and off-targets of CRISPR gRNA spacer sequences using bowtie. 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. Both DNA- and RNA-targeting nucleases are supported.

The software uses the copy number segments from a text file and identifies all chromosome arms that are globally altered and computes various genome-wide scores. The following HRD scores (characteristic of BRCA-mutated cancers) are included: LST, HR-LOH, nLST and gLOH. the package is tailored for the ThermoFisher Oncoscan assay analyzed with their Chromosome Alteration Suite (ChAS) but can be adapted to any input.