Drylab
Use Cases

Run One Analysis on Many Files

Mention a folder and get one result per sample plus a summary table, tested on two files before the full run.

When every sample is a separate file, you don't need to ask once per file. Mention the folder, describe the analysis once, and Drylab runs it on every file and collects the results in one table.

When to use it

  • A sequencing run gave you one FASTQ file (or pair) per sample, and you want QC for each one.
  • You have one flow cytometry file, image or plate-reader export per well, condition or time point.
  • You have many CSV exports with the same columns and want them merged into one table.
  • You want the same figure for every sample, with the same axes and colors, so they're easy to compare.

Prompt examples

Type @o and part of the folder name, then pick the folder. It becomes a chip in your message.

FASTQ QC per sample:

@o fastq_run_01 Run FastQC on every FASTQ file in this folder. Test on 2 files first, then run all. Give me one summary table with the sample name, read count, mean quality and percent GC, one row per sample, and flag samples with mean quality below 30.

Flow cytometry per sample:

@o fcs_run_01 For each FCS file in this folder, report the event count and the median FSC-A and SSC-A. Test on 2 files first, then run all. Put one row per file in a single table and make one scatter plot of FSC-A vs SSC-A per file, with the same axis limits.

Image measurements per field:

@o nuclei_dapi Count the nuclei in every TIFF image in this folder and measure their mean area. Test the segmentation on 2 images first and show me the masks before you run all. Save one row per image in a summary table.

Merge many CSVs:

@o plate_reads Merge every CSV in this folder into one long table with a sample column taken from the file name. Check that all files have the same columns first, and list any file that doesn't.

Summary per file:

@o plate_reads For each CSV in this folder, compute the mean, median and max od600 and the number of wells above 0.8. Test your code on 2 files first, then run all. Save one summary row per file in a single table and make one bar plot of the mean od600 per sample.

Steps

  1. Put all the files for the batch in one folder of your project. To upload a whole folder, click + (Upload files or mention resources) in the chatbox and choose Upload folder. It goes to the project's Project Files folder. See Upload Files.
  2. In the chatbox, type @o (with a space) and part of the folder name. The Folders tab lists matching folders with their path. Pick the row whose path ends in Project Files/<your folder>, not a chat's folder under History.
  3. Pick the folder. It becomes a chip in your message.
  4. Describe the analysis once, then say what you want back: one result per file and one summary table. Add "test on 2 files first, then run all".
  5. Send it. Drylab checks the files, runs the analysis on two of them, then on all of them, and answers with the summary table and the files it saved.
  1. Copy the folder with your files into your project folder with Finder, File Explorer or your file manager. It shows up in the File tree on its own. Nothing is uploaded. See Work with Local Files.
  2. In the chatbox, type @o (with a space) and part of the folder name. In the Desktop App, @ searches your whole project folder.
  3. Pick the folder. It becomes a chip in your message.
  4. Describe the analysis once, then say what you want back: one result per file and one summary table. Add "test on 2 files first, then run all".
  5. Send it. The results are saved in this chat's folder under History, inside your project folder.

If your request matches one of Drylab's structured workflows, a Structured workflow detected card may ask how to proceed. Click No workflow to run your request as written. See Plans and Questions.

Tips

  • Test on two files first. A problem with the file format (a comment line, a different column name) then shows up in seconds instead of after the whole batch.
  • Say how a file maps to a sample, for example "the sample name is the file name without .csv", or point to a sample sheet with @samples.csv.
  • Ask for one summary table, one row per file, plus the per-file results. It's the table you'll actually read.
  • Keep the batch in its own folder. A folder that also holds old or unrelated files gets them included too.

Continue from an Earlier Result

Pick up from a table or object saved by an earlier chat instead of running the whole analysis again.

Run a Guided Workflow

Let a step-by-step workflow run a standard analysis and ask you at each key decision.

On this page

When to use it
Prompt examples
Steps
Tips
Related