Create an interface that, given a plate, provides its ancestry and descendency

[josenavas]

i.e. which plates/process comes from and which plates/process have been generated from it.

[AmandaBirmingham]

I suggest that a sensible terminology for this is "plate lineage".

[tanaes]

Specific questions one will likely need to answer:

• Given a sample, what plate does it appear in?
• Given a library plate, what DNA plates did it pull from?
• Given a sequencing run, what DNA plates did it pull from?
• Given a Qiita study, what sequencing runs are associated with the study?
• Given a DNA plate, was a library plate made from it?
• Given a library plate, what are the 4 DNA plate barcodes associated? I need to get them out of a freezer

[tanaes]

A rough sketch from @ElDeveloper and my online whiteboard session... think is that if a plate ID is queried, all the incoming edges could be retrieved by searching the plates of origin of thee ingoing compositions, and so on up to original plated samples; and all the outgoing edges could similarly be retrieved, but you would not need to "turn the corner" and retrieve other ingoing compositions to downstream plates.

[AmandaBirmingham]

Just so this doesn't get lost in old email archives, here's the pseudocode I taped out back in January for what some of the necessary db queries would look like:

# Note: throughout this pseudocode, "subtype composition" means one of the specific composition types like sample_composition,
# gdna_composition, library_prep_shotgun_composition, etc.
# "subtype process" means one of the two *_process_data tables (gdna_extraction_process_data or library_prep_16s_process_data);
# if these tables are removed from the db schema, then the functionality of Get info for processes  becomes a little simpler.

Get ancestor processes (plate id):

              get all of plate’s wells from well by plate_id
               for each well
                               get its container_id from well
                               get the latest_upstream_process_id for that container_id from container
                               add latest_upstream_process_id to a list of unique process ids                   

                               get the composition_id and composition_type_id for this upstream_process_id and container_id from composition
                               look up the composition_id in the appropriate subtype composition table

                               <recurse up through input compositions>:
                               get all the input composition ids for the subtype composition (e.g., sample_composition_id is an input to gdna_composition, gdna_composition_id is an input to library_prep_16s_composition, etc)
                               for each of these input composition ids
                                               look up the input composition id in the appropriate subtype composition table (e.g., if the input is a gdna_composition_id, you look it up in gdna_composition)
                                               get the composition_id for that subtype composition
                                               get the upstream_process_id of this composition from the composition table by composition_id
                                               add process id to the list of unique process ids
                                               <recurse up through input compositions>

               # at the end of this, we have a list of all the process ids that have contributed to this plate
               # either directly or indirectly

               call Get info for processes with that list as input, and return results

Get descendants processes (plate id):

                get all of plate’s wells from well by plate_id
                for each well
                                get its container_id from well
                                get all the composition_ids for that container_id, and their associated composition_type_ids, from composition
                                for each composition_id
                                                look up the composition id_in the relevant subtype composition table
                                                get the subtype composition id
                                                <recurse down through input compositions>:
                                                look up the subtype composition id in each of the subtype composition tables that takes that kind of id as an input (e.g., if you've got a gdna_composition_id, look it up in the gdna_composition_id field of normalized_gdna_composition and in the gdna_composition_id field of library_prep_16s_composition)
                                                for record found
                                                                get the subtype composition id and the composition_id for the subtype composition having the looked-up value as an input
                                                                look up the composition_id in the composition table and get its upstream_process_id
                                                                add that process id to list of unique process ids
                                                                <recurse down through input compositions>
 
                # at the end of this, we have a list of all the processes to which material from this plate was an input
                # either directly or indirectly
 
                call Get info for processes with that list as input, and return results
 

Get info for processes(process_ids_list)

                for each unique process id in process_ids_list in ascending order of run date
                                look up process_id in process
                                get the process_type_id for this process
                                look up the process_id in the subtype process table
                                if the subtype is NOT gdna_extraction or library_prep_16s
                                                gather all fields that aren't the primary key or the process_id foreign key--these are the settings
                                else
                                                get the subtype process id
                                                get all the records in the *_process_data table for this process subtype having this subtype process id
                                                for each of them
                                                                gather all fields that aren't the primary key--these are the settings
 
                # return the list of process/setting info, in ascending order of run date
 

[tanaes]

Awesome! What would it mean to remove the *_process_data tables?

also attn: @ElDeveloper

[AmandaBirmingham]

@tanaes Actually, the *_process_data tables have already been removed (the email I copied was written before that fix, sorry) so the true sql queries will be a little simpler than the ones above.

[tanaes]

Awesome, thanks!

…

On Fri, Apr 13, 2018 at 11:35 AM Amanda Birmingham ***@***.***> wrote: @tanaes <https://github.com/tanaes> Actually, the *_process_data tables have already been removed (the email I copied was written before that fix, sorry) so the true sql queries will be a little simpler than the ones above. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#77 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AH6JAJbMa2oCiDtXLKwThunC4I5VbKsJks5toO_fgaJpZM4RkhIZ> .

[tanaes]

@qiyunzhu @mortonjt A minimal solution to this issue would be to:

1. take a plate ID input, e.g. from the Plate List interface
2. identify all the Composition elements within that plate
3. for each Composition:

• follow it 'downstream' in the database to find all Composition objects which inherit that composition
• identify all the plate IDs in which those downstream compositions were resident
• return a table showing all of those plate IDs.

4. repeat for 'upstream' compositions

[charles-cowart]

@AmandaBirmingham Wondering if the wet-lab needs this feature for full launch, or if it can be added later.

[AmandaBirmingham]

@charles-cowart no one from the actual wet lab has ever asked me for it/about it, so I'm not convinced of its importance.

[AmandaBirmingham]

Note per discussion with @charles-cowart and @wasade today: If/when we implement this functionality (particularly the descent functionality), we would then have the necessary logic to check whether a plate can be renamed (plates that have no descendants of any of their wells can safely be renamed). However, note that @wasade notes that multiple users could be using LabControl at the same time, so descendants could be changing on the fly!