broadinstitute/scalable_analytics

Public collaboration of Scalable Single Cell Analytics

scalable_analytics

The code in this repository is designed for use with single-cell RNAseq data to help determine the cell types present in the dataset.

Perform the following four steps to obtain results:

1. Load data
2. Perform quality control
3. Cluster filtered data
4. Compute differential expresssion among the clusters

The analyses here are based on those in https://github.com/broadinstitute/BipolarCell2016 and https://github.com/broadinstitute/single_cell_analysis ported to tools and techniques available (but not limited to) Google Cloud Platform.

• All steps occur in the cloud.
• Data loading makes use of Docker, and dsub via Compute Engine for batch processing.
• The analyses make use of:
  • Standard SQL via BigQuery
  • Apache Beam via Dataflow
  • TensorFlow via Cloud Machine Learning Engine
• We suggest working through the introductory materials for each tool before working with the code in this repository.