could not find function "KotliarovPBMCData"

[Yale Liu]

Hi singleR team,

Thanks for such great annotation package.

Recently, I want to use the KotliarovPBMCData() to annotate my datasets. I checked I had this reference data, but when I load it, I got error:

> sce <- KotliarovPBMCData()
Error in KotliarovPBMCData() : 
  could not find function "KotliarovPBMCData"

Can you help me with it?

Thanks,
Yale

[Aaron Lun]

That's a function from the scRNAseq package.

[Yale Liu]

Thanks, I see. After I ran the following code, it works.
sce <- scRNAseq::KotliarovPBMCData()

Thanks again!
Yale

[Yale Liu]

Hi LTLA,@LTLA

One more question: After I loaded the reference datasets, I found they have no label.main. Whether it means I can not use it as a reference to annotate my own datasets?

colnames(sce@colData)
[1] "nGene" "nUMI"
[3] "orig.ident" "pctMT"
[5] "barcode_check" "tenx_lane"
[7] "cohort" "batch"
[9] "hash_maxID" "hash_secondID"
[11] "hash_margin" "hto_classification"
[13] "hto_classification_global" "hash_ID"
[15] "adjmfc.time" "DEMUXLET.RD.PASS"
[17] "DEMUXLET.N.SNP" "DMX_GLOBAL_BEST"
[19] "DEMUXLET.BARCODE" "sample"
[21] "sampleid" "joint_classification_global"
[23] "dmx_hto_match" "timepoint"

Thanks again!
Yale

[Aaron Lun]

I can not use it as a reference to annotate my own datasets?

That is correct. The scRNAseq package just serves up the author-provided dataset "as-is". We don't do any re-analysis to label the cells, so if the authors didn't label it - or, at least, the labels weren't publicly available - then you're out of luck.

Sometimes the author-provided labels are there but are hidden away deep in the supplementary materials. If you really care about this dataset, you could try digging around in the authors' paper to see if they provide labels in some mystery file or on their website. We didn't spend too much time trying to pull this information out, because it's a pain to do.

I mean, I guess I could label the cells myself, but my PBMC knowledge is superficial, so I wouldn't think that my labels would be worth the metaphorical paper it's printed on.

P.S. It is bad practice to use @ in your analysis code. The class internals are not guaranteed to be stable, so if the structure of the class changes across versions, your code will not work. Use the getter function colData(sce) instead.