Landmark gene selection in L1000.

Question

[kim]

Hello, I'm a beginner in bioinformatics and I have some question about cmap, precisely, regarding Landmark gene selection. I am grateful for all of your responses. here are a few questions. It's a question about paper "A Next Generation Connectivity Map: L1000 Platform and the First 1,000,000 Profiles. Subramanian A, Narayan R et al

1. I want to learn about the Affymetrix HGU133A microarray gene expression dataset that was exactly used when choosing the 978 landmark genes. I've tried to find some data, but there are so many GPLs with Affymetrix HGU133A microarray, and their number of rows is not 12063 as mentioned in the published paper. I think the data may have been updated, or perhaps I'm misunderstanding something.
2. I'm curious about the PCA method in paper. What software tool did you use to obtain the 386 components?

I'm researching how to improve inference, so I want to understand how you reduced the 12063 genes to 978 genes. Thank you for reading, and thank you for your help. I'll be waiting for your response!

[GenoMax]

Paper referred to in question: https://pubmed.ncbi.nlm.nih.gov/29195078/

so I want to understand how you reduced the 12063 genes to 978 genes

From the paper:

Gene space summary

The L1000 assay directly measures 978 genes and infers 11,350 more, for a total of 12,328 genes. Of the 11,350 inferred genes, 9,196 are considered well inferred, based on the analysis described above. All datasets are provided in the full 12,328 gene space.

Table S3 supposedly has this information.

I've tried to find some data

Data can be found at NCBI GEO : https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE92743

[kim]

thank you for reply, but i want some information for what microarray data are selected for L1000. thank you!

[GenoMax]

Link for the data is included above.