Data Analysis
Covers the whole arc from getting data to drawing a conclusion, in six groups: bioinformatics, data and computing, research tools, statistics and plotting, biology databases, and chemistry tools. The biology databases run from protein to pathway: UniProt and the Gene Ontology for sequences and function, RCSB PDB for 3D structures, the Human Protein Atlas for expression, KEGG and Reactome for pathways, Metascape for enrichment of a gene list, and PRIDE for finding and downloading raw mass-spectrometry data — EMBL-EBI is the umbrella portal for all of them. The bioinformatics group is the analysis itself: NCBI and BLAST for sequences, Ensembl and the UCSC Genome Browser for genomes, STRING, GEPIA and GeneCards for interactions and expression, DAVID and Enrichr for enrichment, cBioPortal for cancer genomics; take scripting and method questions to Biostars, where most have been answered already. Data and computing covers general-purpose maths alongside public datasets — Wolfram Alpha, GeoGebra, then Kaggle, Google Dataset Search, Data Commons, Our World in Data, China's NBS, and ScienceDB for DOIs on Chinese scientific data. Research tools handles the literature itself: Zotero for references, Overleaf for typesetting, Elicit, Consensus, Connected Papers and Research Rabbit for review and citation graphs. Chemistry tools is the group kept open on a lab bench: PubChem for structures and properties, plus the mass-spec and gel calculators — molecular mass, peptide mass, fragment ions, gel recipes — with MDCalc for medical formulas. The statistics and plotting group is the one link in this chain about getting the result *right* rather than getting it at all, and it is where most people actually stall. When you cannot tell which test applies, how large a sample needs to be, or how to read a p-value, the explainers at Statistics by Jim and Cross Validated are faster than any manual; for doing the analysis, R with Bioconductor is the academic and bioinformatics mainstream, JASP and jamovi are free with SPSS-style interfaces and near-zero migration cost, and among commercial packages GraphPad Prism is the biomedical convention, Origin leans physical-science and Stata leans econometrics; on the Python side pandas handles the data, matplotlib is the substrate and seaborn turns it into statistical plots quickly. The most practical layer is plotting by reverse lookup — The R Graph Gallery and Statology let you start from the chart you need and adapt the code; ggplot2 deserves learning properly once, because its grammar of graphics beats a point-and-click interface the moment you need to redraw. Look databases up by accession or PDB ID rather than by keyword — it is both faster and more accurate.
