How to open Sid Sijbrandij's osteosarcoma multi-omic dataset in a Jupyter notebook
Start with manifest.txt at the bucket root, which lists all 397,036 objects, then read individual tables with pandas. In a Scigantic notebook the 38 TB bucket is mounted at /mnt/archive, so one file can be opened without copying the rest.
The code on this page runs in a Scigantic hosted notebook, where the dataset is mounted read-only at /mnt/archive. The same path is in the SCIGANTIC_MOUNT_PATH environment variable and linked from ~/archive. The public source is s3://sid-sijbrandij-osteosarc-dataset (AWS us-west-2, public, no account needed).
Open Sid Sijbrandij's osteosarcoma dataset on SciganticOpen the sample notebook: A tumor mapped across time
Code
Count objects per folder and file type from the manifest
# Orient yourself with the manifest before touching 38 TB.
import pandas as pd, collections, urllib.parse
m = open('/mnt/archive/manifest.txt').read().splitlines() # 397,036 S3 URLs (2026-09-29 manifest)
top = collections.Counter(urllib.parse.urlparse(u).path.lstrip('/').split('/')[0] for u in m)
print(pd.Series(top).sort_values(ascending=False).head(20)) # objects per top-level prefix
ext = collections.Counter(u.rsplit('.',1)[-1].lower() for u in m if '.' in u.rsplit('/',1)[-1])
print(pd.Series(ext).sort_values(ascending=False).head(15)) # file-type distribution
The starter cell from this dataset's schema card. It ran without error on 2026-11-14 in the Scigantic notebook image against the mounted data.
Read Xenium spatial transcript counts
import os
import pandas as pd
df = pd.read_csv(os.path.join(os.environ['SCIGANTIC_MOUNT_PATH'], 'xenium/cartoscope/output/T0_B3/sge/transcripts.unsorted.tsv.gz'), nrows=1000)
df.head()
Run on 2026-11-14 in the Scigantic notebook image against the mounted data. It returned: CSV/TSV: columns ['X', 'Y', 'gene', 'count'], dtypes ['float64', 'float64', 'object', 'int64'], first 1000 rows read
What the files are
One patient's osteosarcoma, sampled across about five timepoints from December 2022 to April 2025 and released as public domain data by the Rare Cancer Research Foundation.
It spans bulk DNA and RNA sequencing, single-cell RNA-seq, Oxford Nanopore long reads, 10x Xenium spatial transcriptomics, multiplex pathology imaging, clinical CT, MR and PET studies as DICOM, flow cytometry and neoantigen predictions.
How the files are laid out
| Path under the mount | What it holds |
|---|---|
| manifest.txt, md5sums.txt | Every object as an S3 HTTPS URL (60.5 MB), and checksums. |
| genomics/, genomics_reprocessing/ | Bulk DNA and RNA from the nf-core/sarek pipeline: aligned BAM and CRAM files by date, vendor and assay. |
| hudson_lab/ | PBMC single-cell RNA-seq (Cell Ranger output, FASTQ, BAM), flow cytometry .fcs files. |
| xenium/ | 10x Xenium spatial transcriptomics runs. xenium/cartoscope/output/T0_B3/sge/transcripts.unsorted.tsv.gz has columns X, Y, gene, count. |
| ucsf/, ONT/ | Oxford Nanopore long-read data, including raw .pod5 signal. |
| dicom/ | 137 dated clinical imaging study folders (CT, MR, XR, PET). |
| hms_spatial/ | Multiplex pathology as Minerva deep-zoom JPEG tiles, plus an 8.28 GB OME-TIFF. |
Size, format and license
| Format | BAM, CRAM, FASTQ, POD5, DICOM, OME-TIFF, CSV and TSV, among others |
|---|---|
| Size | 397,036 objects, about 38 TB |
| License | CC0 1.0 |
| Provider | Rare Cancer Research Foundation, on AWS Open Data |
| Public source | s3://sid-sijbrandij-osteosarc-dataset (AWS us-west-2, public, no account needed) |
| Notebook image | Genomics (pandas, pysam and scanpy preinstalled) |
Things to know before you read it
- Over half the objects are image-pyramid tiles (218,021
.jpg) and a third are DICOM slices, so object counts say little about sample counts. - Use the manifest to find files. Do not walk the mount.
dicom/andDICOM/are different folders.
Run it
Start a notebook from the Sid Sijbrandij's osteosarcoma dataset archive page. The data is mounted when the notebook opens, so the code above runs as written and nothing is downloaded first. For a longer worked example, open the sample notebook A tumor mapped across time.
Related guides
- How to query 1000 Genomes VCF files on S3 from a Jupyter notebook
- How to open Tahoe-100M single-cell perturbation data in a Jupyter notebook
- How to load CMMD mammography DICOM files and clinical data in a Jupyter notebook
- How to open CAMELYON lymph node whole-slide images from S3 in a Jupyter notebook
Last checked 2026-11-21. All dataset guides.