DPL-899 007c Phase 2 discovery

[KatyTaylor]

Description
Create stories for 007c Phase 2, including:

• which are MVP
• size estimate (S, M, L)

Who the primary contacts are for this work
Katy

Knowledge or Stake holders
Lesley, Abby

[KatyTaylor]

Stories for next phase of pipeline - mostly in pipeline chronological order:

To do:

• Record testing story / plan - compound sample creation.

Pre-requisites:

1. DPL-968 Research - intermediate plates as input plates - DONE
2. DPL-967 Research - chaining submissions - DONE
3. DPL-1011 Re-use of tube racks - enhancement
4. DPL-969 Support for "Chromium chip" labware - Phase 2a
5. DPL-1008 Basic pipeline set up - donor pooling and cDNA prep - Phase 2a

Thawing and pooling: - Phase 2a

1. DPL-913 Submit banked PBMC tubes for cDNA prep
2. DPL-914 Create tube rack containing banked PBMC tubes
3. DPL-1012 Transfer PBMCs from tube rack to plate
   • Bed verification details TBC
4. DPL-915 Count thawed cells
5. DPL-916 PBMC donor pooling - create plate
6. DPL-1024 PBMC donor pooling - driver file
7. DPL-1025 PBMC donor pooling - bed verification

cDNA prep: - Phase 2a

1. DPL-823 Import PBMC pool plates into Sequencescape
2. DPL-917 Submit plate of PBMC pools for cDNA prep
3. DPL-919 Transfer PBMC pools to 10X chip
4. DPL-920 Transfer from 10X chip to GEMs plate, in duplicate
5. DPL-921 Stamp to cDNA PCR plate - cDNA appears at this step - can freeze down here - or maybe the next plate if needs to be cleaned up first.
6. DPL-922 Consolidate into cDNA PCR XP plate
   • Bed numbers for bed verification not finalised (waiting on R&D but not urgent as story could be mostly done anyway)

Library prep: - Phase 2b

1. DPL-970 Import cDNA plates into Sequencescape
   • Plan to meet with CellGen to warn them about this change to process (PMs organising)
2. DPL-971 Create 'HT' version of 'Bespoke Chromium 5 prime' pipelines
3. DPL-924 10X PCR binning strategies change

Improvements:

1. DPL-972 View samples within a compound sample - enhancement

[KatyTaylor]

Idea of how pipeline config could be:

# Thawing, pooling, GEM generation and cDNA Prep
# with an extra entry point half way down.
scRNA Core cDNA Prep:
  filters:
    request_type_key: limber_scrna_core_cDNA_prep
    library_type: Chromium single cell 5 prime HT v2
  relationships:
    LRC Bank Seq: LRC PBMC Thaw
    LRC Bank Spare: LRC PBMC Thaw
    LRC PBMC Thaw: LRC PBMC Pools
    LRC PBMC Pools: LRC HT 5p Chip
    LRC PBMC Pools Fac: LRC HT 5p Chip # some plates come in here, direct from faculty
    LRC HT 5p Chip: LRC HT 5p GEMs
    LRC HT 5p GEMs: LRC HT 5p cDNA PCR
    LRC HT 5p cDNA PCR: LRC HT 5p cDNA PCR XP

We could do the above version, as they have decided they won't be combining plates thawed and pooled in SeqOps with plates thawed and pooled in faculty. However, if they changed their mind in future, it might be easier if we did it as 2 separate pipelines, as below. However, this makes it more complicated as we'd have more different requests, request types etc, and would have to chain the requests together somehow so you had the option of submitting tubes for both pipelines at once.

# Thawing PBMCs and pooling donor samples
scRNA Core Donor Pooling:
  pipeline_group: scRNA Core cDNA Prep
  filters:
    request_type_key: limber_scrna_core_donor_pooling
  relationships:
    LRC Bank Seq: LRC PBMC Thaw
    LRC Bank Spare: LRC PBMC Thaw
    LRC PBMC Thaw: LRC PBMC Pools

# GEM generation and cDNA Prep
scRNA Core cDNA Prep:
  pipeline_group: scRNA Core cDNA Prep
  filters:
    request_type_key: limber_scrna_core_cDNA_prep
    library_type: Chromium single cell 5 prime HT v2
  relationships:
    LRC PBMC Pools: LRC HT 5p Chip
    LRC PBMC Pools Fac: LRC HT 5p Chip
    LRC HT 5p Chip: LRC HT 5p GEMs
    LRC HT 5p GEMs: LRC HT 5p cDNA PCR
    LRC HT 5p cDNA PCR: LRC HT 5p cDNA PCR XP

Pipeline diagrams here

[KatyTaylor]

Categorised list of stories, post GEM-X decision:

Phase 1 - high priority:

• Y24-037 Cell count spot-checking #1653 - Done
• Y24-096 Make barcode shorter - LRC Blood Aliquot #1707 - Done
• Y24-123 Cell counting - QC labels - quick fix #1729 - Done

There are also several Phase 1 enhancement stories in the backlog.

Phase 2a - main pipeline flow:

• DPL-919 Transfer PBMC pools to 10X chip #1392 - Done
• DPL-920 Transfer from 10X chip to GEMs plate #1393 - Done
• Y24-171 - Driver file: PBMC pools --> 10x chip #1753 - Done
• Cell count post pooling (waiting on R&D for more info)
• DPL-980 Import PBMC tubes into Sequencescape sequencescape#3951 - UAT
• Y24-169 - Allow faculty input PBMC tubes down pipeline sequencescape#4159 - UAT

Phase 2a - essential changes:

• Y24-083 Cell counting driver file (LRC PBMC Defrost PBS plate) #1688 - Done
• Y24-007 Updates to PBMC donor pooling - create plate #1611 (On Hold)
• DPL-1025 PBMC donor pooling - bed verification #1504 - Done
• Y24-177 - Modify Aggregation submission #1767 - In Progress
• Y24-178 - Modify cDNA Prep submission #1768 - UAT
• Y24-214 - Modify cDNA Prep 'faculty input' submission #1813 - To Do
• DPL-918 Extra cell counting step - pooled PBMCs #1391 - To Do

Phase 2b - main pipeline flow:

• DPL-971 Create 007c version of 'Bespoke Chromium 5 prime' pipelines #1444 - UAT
• DPL-970 Import cDNA plates into Sequencescape sequencescape#3937 - To do

[KatyTaylor]

Writing notes to help decision around the donor pooling step - 1 source plate or 2:

1 source plate

Argument for doing this is that the first iteration of the pipeline will not require more than 1 source plate. The UX will be better as the user will not be asked to scan a plate into an extra screen.

Stories to do:

• Y24-090 Donor pooling - only one source plate #1706

2 source plates

Argument for doing this is that when the Cardinal main phase goes ahead, it may require 2 source plates here, and we have already built the interstitial page for this so wouldn't waste this work.

Stories to do:

• Y24-048 Scanned plate must match previous page - PBMC donor pooling #1665
• Y24-047 Restrict plate purpose - PBMC donor pooling [Deprecated] #1664

I suggest selecting the '1 source plate' option, because
a) the user experience will be better
b) there's extra work to do to implement 2 source plates properly anyway
c) we won't lose the work we've done on the 2 source plates option, as it will be in git
d) Cardinal requirements are quite a way off and uncertain

Update: discussed with Abdullah and we agreed to go with the '1 source plate' option for now.