2022-Kleftogiannis et al-preprint

[mleipold]

"Identifying predictors of survival in patients with leukemia using single-cell mass cytometry and machine learning"
Dimitrios Kleftogiannis, Benedicte Sjo Tislevoll, Monica Hellesøy, Stein-Erik Gullaksen, Nisha van der Meer, Emmanuel Griessinger, Inga K. F. Motzfeldt, Oda Fagerholt, Andrea Lenartova, Yngvar Fløisand, Jan Jacob Schuringa, Bjørn Tore Gjertsen, Inge Jonassen
bioRxiv, posted August 15, 2022
https://doi.org/10.1101/2022.08.13.503587

- will update when/if this is peer-reviewed and published

- "The CyTOF datasets presented in this study are available at https://flowrepository.org/with ID: FR-FCM-Z5YT, and more details can be provided upon reasonable request."
- "Sorry, you do not have access to the specified experiment."

- also reuses 2015-Levine et al-Cell

[sgranjeaud]

Now available http://flowrepository.org/id/FR-FCM-Z5YT
The key to link each file to the patient ID in the manuscript is added as an attachment.

32 patients with time point 0h, 4h, 24h
7 healthy donor samples

The data analysis framework has been implemented using the R language. The codes along with a tutorial is available in GitHub https://github.com/dkleftogi/singleCellClassification.

Data pre-processing and normalisation of all CyTOF samples is described in Tislevoll et al. (2023). Nat Commun 14, 115.

All fcs files have been compensated by the Catalyst pipeline, then normalized with the CytoNorm (without clustering), which is described in Tislevoll, as cited below.

After the acquisition, the collected data was normalized to EQ bead standard67. Cells were gated by DNA-Ir191 versus event length, followed by the exclusion of cell doublets through stringent gating using the two DNA stain channels (Ir191/Ir193). Files were debarcoded using a single-cell debarcode algorithm (https://github.com/nolanlab/single-cell-debarcoder). All data were arcsinh-transformed (cofactor 5). To limit the possibility of spillover between antibodies channels, the data were compensated according to the CATALYST (Cytometry Data analysis tools) pipeline. https://doi.org/10.18129/B9.bioc.CATALYST. The study was performed before the compensation method was available, and we, therefore, used beads created in a later experiment for compensation. The beads were stained with other antibodies conjugated to the same metal isotopes. We assume the difference in metal abundance between the two experiments is negligible, and that the potential difference in metal abundance between the two experiments will be outweighed by the advantage of compensating for the data.
Subsequently, based on the reference samples, all samples were standardized across the seven barcodepools using quantile normalization. Barcodepool 7 was selected as the normalization standard as it had the best separation of populations and the least background staining when the median intensity of all surface markers was compared between barcodes. Standardization was performed following the CytoNorm approach, but without clustering, in 101 quantiles69. In short, we calculated 0–100% quantiles for the reference samples in each barcode. For each barcode, we then calculated the piecewise linear function whose value at the ith quantile is the ith quantile of the reference barcode. Subsequently, the piecewise linear functions were applied to all non-reference samples.

[mleipold]

Hi Samuel,

I still don't have access to Z5YT. Even going in directly to FR and using the last 4 characters to replace another link (to avoid ASCII character issues around dashes in printed links), I still get "Sorry, you do not have access to the specified experiment."


Mike

[mleipold]

Hi all,

After several emails back and forth, Samuel and I appear to have figured out the issue.

The Z5YT dataset link doesn't work, normally. The authors apparently haven't switched it to public.


The reason that Samuel could access it before was because of the FR reviewer link in the Tislevoll et al paper: viewtopic.php?f=10&t=2985

If you follow the reviewer link, that appears to give your browser a cookie that then allows the Z5YT link to work directly if you later try it.


Mike

[mleipold]

Hi all,

Z5YT is now accessible. Thanks to the authors for making the switch.


Mike

[mleipold]

Now published:

iScience. 2024, 27, 110261
https://doi.org/10.1016/j.isci.2024.110261

- Zenodo link is analysis but no new data; reuses:

FR-FCM-ZYTT: 2019-Abdelaal et al-Cytometry A: AML: actually 2015-Levine et al-Cell; BMMC: 2011-Bendall et al-Science?; PANORAMA: actually 2016-Samusik et al-Nat Methods
FR-FCM-Z5YT: 2023-Tislevoll et al-Nat Commun