Bioinformatics Scientist II
Connectivity Map,Golub Lab
Cancer Program
Computational Biologist
Connectivity Map,Golub Lab
Cancer Program
follow these slides at
http://coreyflynn.github.io/SoapBox/CMAP_hands_onCell lysates are scanned in 384 well plates in the L1000 assay
1000 gene measurements are decovoluted from 500 bead colors
Scale profiles based on expected levels of invariant gene sets and adjust to expected distribution
Use regression models to impute the rest of the transcriptome
Compute Z-scores to determine differential expression
Average between replicates weighted by agreement
Integrate across knockdown signatures targeting the same gene to generate consensus gene signatures
Assemble signatures into a unified matrix
Compare one signature to all other signatures in the database looking for connections
Summarize the results of a query to focus on the most reliable connections
Look for consistency in related signatures
Get metadata for signatures
Make subsets of the data matrix