Lab Notebook & Data Management SOP

Record-keeping and data-handling standards for protein chemistry and cryo-EM/cryo-ET work. The goal is that any result can be traced — construct to buffer to grid to map to model — by someone who wasn't in the room.

1. Electronic lab notebook (ELN)

• Log every experiment in the shared ELN the same day it's run — expression, purification, grid session, and screening notes alike. A result not logged same-day is treated as undocumented.
• Each entry gets a unique experiment ID (see naming convention, Section 2) referenced in every downstream file, plot, and folder derived from it.
• Record what actually happened, not just the plan: deviations from protocol, unexpected observations, and failed conditions are as valuable as successes and must be entered, not omitted.
• Paste in raw instrument output (SEC traces, gels, DLS, screening montages) rather than describing it from memory — attach the source file alongside the note.
• Sign off entries weekly; a lab member's notebook should be legible and auditable by their PI or a collaborator without a walkthrough.

2. Naming & folder conventions

• Use a single ID scheme across notebook, files, and folders: YYYYMMDD_construct_step_initials (e.g. 20260712_ATP5F1B_grid03_JD) — never rely on a project nickname alone.
• Mirror the physical pipeline in folder structure: construct → purification batch → grid session → data collection session → processing run. Don't flatten everything into one directory per person.
• Never overwrite a prior processing run's output directory; append a version suffix and keep a short changelog of what changed between versions and why.
• Keep raw data (movies, tilt series, gels, traces) physically or logically separate from derived/processed data, and never edit raw data in place.

3. Raw data handling & storage

• Offload raw movies/tilt series from the microscope collection PC to networked storage the same day, then verify transfer integrity (checksum or file-count/size match) before deleting the local copy.
• Cryo-EM/cryo-ET raw datasets are large and irreplaceable — maintain at least two copies in physically distinct locations (e.g. primary networked storage + institutional backup or cold archive); a single-copy dataset is considered at-risk.
• Apply consistent access permissions per project (read access for collaborators, write access limited to the data owner) rather than shared open folders.
• Capture acquisition metadata (microscope, detector, magnification, pixel size, dose, defocus range, software versions) in a machine-readable log at collection time — retyping it later from memory introduces errors.

4. Processing provenance

• Record software name and version for every processing step (motion correction, CTF estimation, picking, classification, refinement, model building, validation) — reproducing a map or model later requires knowing exactly what produced it.
• Save job parameters/config files alongside outputs, not just in a tool's internal project database — databases get migrated or lost, flat config files don't.
• Keep intermediate maps/models at key decision points (initial reconstruction, post-classification, final refined) rather than only the final result, so a downstream problem can be traced to its origin.
• Version-control scripts and configs used for processing (git or equivalent) so a pipeline can be rerun exactly as it was for a given deposition.

5. Deposition & long-term archiving

• Deposit final maps to EMDB and coordinates to PDB at publication, including full acquisition and processing metadata in the deposition — not just the minimum required fields.
• Where feasible, deposit representative raw movies/tilt series (e.g. via EMPIAR) so the full dataset — not just the final map — remains reproducible by others.
• Retain raw data for the duration required by funder/institutional policy at minimum; treat that as a floor, not a target — most labs keep primary datasets indefinitely given their cost to reproduce.
• Record the notebook entry ID and folder path for every deposited structure in the lab's master project tracker, so a paper's supporting data can be located years later without relying on any one person's memory.