search

HumanCellAtlas / dcp

Data Coordination Platform manifest and integration tests.
3 stars 1 forks source link

DCP Support for Human 10X snRNA-seq data #378

Open kbergin opened 5 years ago

kbergin commented 5 years ago

Human 10X snRNA-seq support.

Why? Initially the interest in snRNAseq came from the need to be able to process large cells (i.e. cancer) or oddly shaped cells (i.e. brain neurons) that would not process properly in microfluidics due to space constraints. In addition, some cells are delicate and don’t do well with the pressure in the microfluidics and some like skin cells are sticky and die if they can’t sense any neighbor cells. The nuclei of these cells do not have these problems. It was discovered that despite only obtaining pre-mRNA from the nucleus and not being able to get any mitochondrial information, for expression purposes sufficient data is attainable from only the nucleus. Although lab techniques have improved and more difficult cell types can now be processed as single cells, some other logistical standard operating procedures continue to make snRNAseq a preferred option. In hospitals, it is standard operating procedure to flash freeze tissues or samples upon collection. This does aid in transport of the tissue to keep it alive, but frozen cells are very difficult if not impossible to process in single cell experiments. However, nuclei have no problem with being frozen and then processed in this manner. Choosing to do snRNAseq gives scientists access to many more samples stored in hospitals and doesn’t require changes or interrupts to hospital standard procedures.

Q2 Scope Pipeline

Later Pipeline Execution Metadata Browser Portal UX Content Matrix Service

kbergin commented 5 years ago

We made some initial efforts into an snRNAseq pipeline for Optimus. The status of that effort is documented here: https://docs.google.com/document/d/1_C_z0Deg9tVNbS9hl-4xEyOI-6Y4alnnyKSk_63uY9g/edit?pli=1

It will be prioritized later.

kbergin commented 5 years ago

@jkaneria @brianraymor Want to throw in any placeholder epics required to support this in the future? Or is it too soon given we're validating the pipeline Q4?

brianraymor commented 5 years ago

I'd prefer to model when it's a priority (per the roadmap) for a specific Release in a quarter.

kbergin commented 5 years ago

Sounds good! @jahilton of note that if this doesn't get onto roadmap, analysis of this data type will not be available for a data release.

brianraymor commented 5 years ago

I believe that the Tim and Jason are working on a mind meld to reconcile the assay roadmap with the long-term roadmap priorities for the next quarter.