Closed KyndallH closed 6 years ago
Thanks Kyndall,
5= tissue in LN2 within 1 hr of death Should #4 be 2-4 hrs after death? Also should we note LN2, -80 C, or -20 C? 3 = specimen frozen (-20 C) and thawed
4 and 5 come from live animals so theoretically would go into LN2 or -80 and never go into a -20 freezer. 3 is a specimen taken from a -20 C freezer and thawed, then put into LN2 or -80.
This is a continuation of https://github.com/ArctosDB/arctos/issues/1373
I don't really like the numbers - I don't think a user is going to have any idea what "1" means. (But whatever, it's not THAT hard to find our code tables and a definition.)
"temperature" could be another part attribute, perhaps via https://github.com/ArctosDB/arctos/issues/1371. That's embedded in "5" but for the rest that seems a separate and critical determination - it's in LN2 now but was in your frost-free freezer for a decade first, etc.
And I'm not sure we need an explicit "unchecked" option - this will be an optional thing, just don't add data if you don't know.
I created the attribute, a new code table, some test values, and added some determinations to http://arctos-test.tacc.utexas.edu/guid/CHAS:Bird:17187
I would prefer we use both text and numbers. Eg "excellent (5)"
On Jan 4, 2018 6:15 PM, "dustymc" notifications@github.com wrote:
This is a continuation of #1373 https://github.com/ArctosDB/arctos/issues/1373
I don't really like the numbers - I don't think a user is going to have any idea what "1" means. (But whatever, it's not THAT hard to find our code tables and a definition.)
"temperature" could be another part attribute, perhaps via #1371 https://github.com/ArctosDB/arctos/issues/1371. That's embedded in "5" but for the rest that seems a separate and critical determination - it's in LN2 now but was in your frost-free freezer for a decade first, etc.
And I'm not sure we need an explicit "unchecked" option - this will be an optional thing, just don't add data if you don't know.
I created the attribute, a new code table, some test values, and added some determinations to http://arctos-test.tacc.utexas.edu/guid/CHAS:Bird: 17187
— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/ArctosDB/arctos/issues/1384#issuecomment-355446628, or mute the thread https://github.com/notifications/unsubscribe-auth/AOH0hPsEOXXNcS7JNLYE3sy7Mf2HJJI-ks5tHXfJgaJpZM4RTd6I .
We need to make this more generic to other tissue preservation methods, e.g., RNALater. How about something like:
5= preserved (frozen, buffer, ethanol) within 1 hr after death 4= preserved (frozen, buffer, ethanol) within 24 hrs after death 3 = preserved frozen and thawed at least once 2 = somewhat rotten 1 = very rotten 0 = unknown/unchecked
Also, what are the guidelines for distinguishing between 1 and 2 - seems pretty subjective. Could we combine those to just 'rotten' ??? If you can define how 'somewhat' vs 'very' then keep the distinction but provide clearer definitions.
I had the same question as Jon about tissues from non-live animals that were stored in -20. Also, #5 and #4 don't always need to be from live animals. We sometimes get in dead specimens that we prepare right away.
And where would you put tissues preserved in etoh? How do you compare a tissue preserved in etoh within 1 hr after death with a tissue preserved in LN2 that has been thawed at least once? Here's a suggested modification, but we need to add etoh-preserved tissues to this vocabulary (e.g., all of the bird tissues being collected in Indonesia are in etoh as a treatment for avian influenza, USDA requirement).
Excellent (5)= frozen (LN2, -80 C) or preserved in RNALater within 1 hr after death Very Good (4)= frozen LN2, -80 C, buffer) or preserved in RNALater within 24 hrs after death Good (3) = frozen (LN2, -80 C) and thawed at least once Fair (2) = frozen (-20) and thawed at least once Poor (1) = rotten Unknown (0) = unknown/unchecked
dead specimens that we prepare right away.
I don't think those can be 4 or 5?? I don't see how you could know where they've been.
I still don't see any need for "we don't know" mixed in with NULL, which is what we'll (initially) have for most tissues and can only be interpreted as "we don't know."
Can we get at "suitability for use" somehow? A separate vocabulary for ethanol and isopropanol and rnalater and lysis buffer and desiccant and...... seems really awkward and doesn't really get at what users want ("gimme squirrel tissues from which I can probably get {whatever}").
I’m out today with a lot on my plate so I can’t jump back in with quality input but would like to. Can we continue on Monday?
Sent from my iPhone
On Jan 5, 2018, at 10:42 AM, dustymc notifications@github.com<mailto:notifications@github.com> wrote:
dead specimens that we prepare right away.
I don't think those can be 4 or 5?? I don't see how you could know where they've been.
I still don't see any need for "we don't know" mixed in with NULL, which is what we'll (initially) have for most tissues and can only be interpreted as "we don't know."
Can we get at "suitability for use" somehow? A separate vocabulary for ethanol and isopropanol and rnalater and lysis buffer and desiccant and...... seems really awkward and doesn't really get at what users want ("gimme squirrel tissues from which I can probably get {whatever}").
— You are receiving this because you were assigned. Reply to this email directly, view it on GitHubhttps://github.com/ArctosDB/arctos/issues/1384#issuecomment-355617612, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AQeng3UT1yrDDDHyQWad4IpML0G6V8xFks5tHl7lgaJpZM4RTd6I.
Yes, I'm home with some bug and can't do much. Monday would be better.
On Jan 5, 2018 10:56 AM, "jldunnum" notifications@github.com wrote:
I’m out today with a lot on my plate so I can’t jump back in with quality input but would like to. Can we continue on Monday?
Sent from my iPhone
On Jan 5, 2018, at 10:42 AM, dustymc <notifications@github.com<mailto: notifications@github.com>> wrote:
dead specimens that we prepare right away.
I don't think those can be 4 or 5?? I don't see how you could know where they've been.
- 5=you killed it and know the history in some detail
- 4=you're running a trapline or etc., and know history to ~24h precision (which is still sort of avoiding the issue - a lemming that spent 24h in a trap at 40F can be used for different things than a shrew that spent 24h in a trap at 100F, etc.)
- 3="doesn't stink"
- 2= ????
- 1="stinks"
I still don't see any need for "we don't know" mixed in with NULL, which is what we'll (initially) have for most tissues and can only be interpreted as "we don't know."
Can we get at "suitability for use" somehow? A separate vocabulary for ethanol and isopropanol and rnalater and lysis buffer and desiccant and...... seems really awkward and doesn't really get at what users want ("gimme squirrel tissues from which I can probably get {whatever}").
— You are receiving this because you were assigned. Reply to this email directly, view it on GitHubhttps://github.com/ ArctosDB/arctos/issues/1384#issuecomment-355617612, or mute the thread< https://github.com/notifications/unsubscribe-auth/ AQeng3UT1yrDDDHyQWad4IpML0G6V8xFks5tHl7lgaJpZM4RTd6I>.
— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/ArctosDB/arctos/issues/1384#issuecomment-355621145, or mute the thread https://github.com/notifications/unsubscribe-auth/AOH0hPxdzMSSxEZHDjRNr1Eeh4CRDxvgks5tHmI3gaJpZM4RTd6I .
Here are my suggestions based on tweaks to what we do at MSB, with the addition of an extra category for RNA quality samples stored in vapor phase nitrogen. We also need to consider storage issues as well as condition at time of collection.
6=Excellent (RNA quality) = tissues flash frozen in liquid nitrogen or preserved in RNALater within 1 hour of death. Cold chain maintained in liquid or vapor-phase nitrogen with no freeze/thaw exposure. Stored at -180C or in RNALater.
5=Excellent (DNA quality) = tissues flash frozen in liquid nitrogen or preserved in 95% ethanol within 1 hour of death, with no subsequent freeze/thaw exposure. Stored at -80C.
4 =Very Good (DNA quality) = tissues frozen at -20C or below or preserved in buffer or ethanol within 4 hours after death; no more than 1 freeze/thaw events. Stored at -80C and below.
3 = Good (DNA quality) = tissues frozen at -20C or below or preserved in buffer or ethanol within 24 hours after death ; no more than 2 freeze thaw events, Tissues stored at -80C or below.
2= Fair (DNA quality) = tissues degraded, potentially exposed to multiple freeze/thaw events or extended periods above 0C. Stored at -20C or below.
1= Poor (DNA quality) = tissues highly degraded, potentially exposed to multiple freeze/thaw events or extended periods above 0C. Stored at -20C or below.
0=unknown/unchecked.
I suggest for 5-3 instead of DNA quality, genomic quality. 2 would be DNA quality. And 1 questionable DNA quality or something that indicates that there is a very good chance, they will not get DNA out of it but they are welcome to try. Most things that I send out that are 1 are hit and miss if they can get DNA out of. They will let me know which samples worked so I can update it in Arctos.
That works!
On Mon, Jan 8, 2018 at 11:36 AM, Kyndall notifications@github.com wrote:
I suggest for 5-3 instead of DNA quality, genomic quality. 2 would be DNA quality. And 1 questionable DNA quality or something that indicates that there is a very good chance, they will not get DNA out of it but they are welcome to try. Most things that I send out that are 1 are hit and miss if they can get DNA out of. They will let me know which samples worked so I can update it in Arctos.
— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/ArctosDB/arctos/issues/1384#issuecomment-356054297, or mute the thread https://github.com/notifications/unsubscribe-auth/AOH0hAWtvo8fS-ofxuFu6Y1hAN8hUyqaks5tImBAgaJpZM4RTd6I .
Hi All Are we going to discuss this at the AWG meeting? Like Dusty, I am not super in favor of having a number code for quality. I'd prefer a pick list if that is possible. Can we talk on Thursday about this?
On Mon, Jan 8, 2018 at 10:36 AM, Kyndall notifications@github.com wrote:
I suggest for 5-3 instead of DNA quality, genomic quality. 2 would be DNA quality. And 1 questionable DNA quality or something that indicates that there is a very good chance, they will not get DNA out of it but they are welcome to try. Most things that I send out that are 1 are hit and miss if they can get DNA out of. They will let me know which samples worked so I can update it in Arctos.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ArctosDB/arctos/issues/1384#issuecomment-356054297, or mute the thread https://github.com/notifications/unsubscribe-auth/AESS8aE_PY9ZbVoOMDbO7qaQUUjr1Eo8ks5tImBAgaJpZM4RTd6I .
-- Carol L. Spencer, Ph.D. Staff Curator of Herpetology & Researcher Museum of Vertebrate Zoology 3101 Valley Life Sciences Building University of California, Berkeley, CA, USA 94720-3160 atrox10@gmail.com or atrox@berkeley.edu 510-643-5778 http://mvz.berkeley.edu/
Whatever the vocabulary, it will be implemented as a pick list.
http://arctos-test.tacc.utexas.edu/guid/CHAS:Bird:17187
Yes, this is on the agenda for Thursday's meeting.
I think we need to make this simpler. This is too complicated for the average collection user, and we don't want it to be too cumbersome where you have to change the quality every time a sample is thawed (or partially thawed?) for subsampling. We need something that's a useful indicator of fitness for use, but also practical from a management perspective.
I still don't know what I would put for a tissue taken from a carcass that has been stored at -20 (anywhere from 1 day to years, highly variable - don't want to have to change the quality the longer the time in such storage). 3 and 4 are hard to distinguish because it's mixing different variables (hours after death and # freeze/thaw events).
Once we figure this out, we should also think/talk about how we would add this to the data entry bulkloader, so the attribute can be assigned at the time of cataloging.
change the quality every time a sample is thawed (or partially thawed?) for subsampling
Isn't that the point?
carcass that has been stored at -20
Shrew? Maybe 2 (can't be better because it wasn't in -80C within 24h). Moose? It's rotten and nasty because they take a while to freeze.
data entry
There's some discussion of alternatives starting about https://github.com/ArctosDB/arctos/issues/1020#issuecomment-329917799
Add past results, somehow
important to include technique if we add these
more attributes:
AWG meeting suggestions to go with
6 - super (RNA quality) 5 - excellent (genomic quality) 4- very good (genomic quality) 3 - good (DNA quality) 2 - fair (possibly DNA quality) 1 - poor (unknown quality)
Is this a requirement for participating in GGBN, that every tissue have a quality rating? And will this become a required attribute for anything deemed "tissue"?
No and no - it's entirely optional.
It is a GGBN field but not required. We put in standardizing tissue quality as part of the proposal they funded. This would be an optional attribute.
On Fri, Jan 12, 2018 at 11:04 AM, Aren Gunderson notifications@github.com wrote:
Is this a requirement for participating in GGBN, that every tissue have a quality rating? And will this become a required attribute for anything deemed "tissue"?
— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/ArctosDB/arctos/issues/1384#issuecomment-357309169, or mute the thread https://github.com/notifications/unsubscribe-auth/AOH0hLWY6QtAvYChn_YONaMpYfP5xBQzks5tJ560gaJpZM4RTd6I .
AWG meeting suggestions to go with
6 - super (RNA quality) 5 - excellent (genomic quality) 4- very good (genomic quality) 3 - good (DNA quality) 2 - fair (possibly DNA quality) 1 - poor (unknown quality)
What's the difference between 5 and 4? I think we need to expand on the definitions, but not be super specific on times.
Here's my suggestion on a more fleshed out version. Dropdown would be the combination of number and rating.
5 (excellent) - RNA quality. Preserved and stored using an RNA stabilization method (e.g., RNAlater buffer). Preservation is typically soon after death. Minimal thawing has occurred.
4 ( very good) - Genomic quality. Preserved and stored using a method that is fit for genomic use (e.g., liquid nitrogen). Preservation is typically soon after death. Minimal thawing has occurred.
3 (good) - High DNA quality. Preserved and stored using a method that is fit for standard, high quality DNA use (e.g., liquid nitrogen, -80 freezer, -20 freezer). Post-mortem preservation times and amount of thawing variable, but likely (or known) to generate high quality DNA with standard methods.
2 (fair) - Low DNA quality. Possible degradation of DNA due to long post-mortem preservation time, repeated or extensive thawing, and/or poor storage conditions. May also have been tested for DNA with low success.
1 (poor) - Low DNA quality. Known degradation of DNA due to long post-mortem preservation time, repeated or extensive thawing, and/or poor storage conditions. May also have been tested for DNA with poor success.
0 (unknown) - No information on quality available.
3 (good) -...-20 freezer
Is that correct?
First search result: https://www.ogt.com/resources/literature/403_dna_storage_and_quality
Storage at –20°C is not recommended as this can result in lower yields.
0 (unknown)
I still don't get it. Someone pulls out a tissue, looks at it, realizes that they have nothing useful to say, and records an attribute asserting that they don't know anything? Functionally, how is "we don't know anything" different than not adding an attribute at all (which can only be interpreted as "we don't know anything")? I'm not suggesting you can't do this, but I do need to know the functional difference between "0" and NULL.
My suggestions in caps. The reality is that "genomic" quality is not necessarily a great term as many of the new genome technologies can do quite well with fragmented DNA. I think this has more to do with degradation (post mortem processing, storage time/temp, and number of freeze thaw, Also, I don't think there's any evidence that RNA buffers are "better" than a tissue that goes into (and stays in) liquid nitrogen.
On Fri, Jan 12, 2018 at 3:50 PM, Carla Cicero notifications@github.com wrote:
AWG meeting suggestions to go with
6 - super (RNA quality)--INTO BUFFER OR LN2 W/IN 10 MINUTES 5 - excellent (genomic quality)--INTO BUFFER OR LN2 WITHIN 1 HOUR 4- very good (genomic quality)--WITHIN 2 HOURS, STORED IN -80C FREEZER 3 - good (DNA quality)--STORED AT -20 FOR SOME PERIOD 2 - fair (possibly DNA quality)--DEGRADATION DUE TO LATE PRESERVATION OR MULTIPLE FREEZE/THAW 1 - poor (unknown quality)
0-UNKNOWN
What's the difference between 5 and 4? I think we need to expand on the definitions, but not be super specific on times.
Here's my suggestion on a more fleshed out version. Dropdown would be the combination of number and rating.
5 (excellent) - RNA quality. Preserved and stored using an RNA stabilization method (e.g., RNAlater buffer) OR LN2. Preservation is typically soon after death. Minimal thawing has occurred.
4 ( very good) - Genomic quality. Preserved and stored using a method that is fit for genomic use (e.g., liquid nitrogen). Preservation is typically soon after death. Minimal thawing has occurred.
3 (good) - High DNA quality. Preserved and stored using a method that is fit for standard, high quality DNA use (e.g., liquid nitrogen, -80 freezer, -20 freezer). Post-mortem preservation times and amount of thawing variable, but likely (or known) to generate high quality DNA with standard methods.
2 (fair) - Low DNA quality. Possible degradation of DNA due to long post-mortem preservation time, repeated or extensive thawing, and/or poor storage conditions. May also have been tested for DNA with low success.
1 (poor) - Low DNA quality. Known degradation of DNA due to long post-mortem preservation time, repeated or extensive thawing, and/or poor storage conditions. May also have been tested for DNA with poor success.
0 (unknown) - No information on quality available.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ArctosDB/arctos/issues/1384#issuecomment-357376186, or mute the thread https://github.com/notifications/unsubscribe-auth/AFx1EvkwLkaRdp_pyWKxyySho1GFVBXVks5tJ-HCgaJpZM4RTd6I .
Doesn't type of buffer matter? RNAlater vs DMSO gives very different qualities long-term. I think we need to be more specific there.
Very good - Would also apply to tissues stored in LN2. Good - Would also apply to tissues stored at -80. Fair vs Poor - There are ones that we think are degraded (tissue looks a little funny when prepping), and ones that we know are degraded because someone left out the vial or the freezer thawed and didn't get noticed for several days. We should distinguish.
I still don't like the distinction between 4 and 5, and think that we can get it to a scale of 1-5.
Finally, we can't always rely on collectors to give us this information. Students should be able to enter quality when they're entering tissue parts during cataloging.
Try again:
5 (excellent) - RNA quality. Preserved and stored using a method that stabilizes RNA (buffer, liquid nitrogen). Preservation within 10 minutes post-mortem. Minimal thawing has occurred.
4 ( very good) - Genomic quality. Preserved and stored using a method that has high fitness for genomic use (buffer, liquid nitrogen). Preservation within 1 hour post-mortem. Minimal thawing has occurred.
3 (good) - High DNA quality. Preserved and stored using a method that is fit for standard, high quality DNA use (liquid nitrogen, -80 freezer, -20 freezer). Preservation may be more than 1 hour post-mortem. Minimal thawing has occurred. Likely (or known) to generate high quality DNA with standard methods.
2 (fair) - Low DNA quality. Possible degradation of DNA due to long post-mortem preservation time, repeated or extensive thawing, and/or poor storage conditions. May also have been sequenced with low success.
1 (poor) - Low DNA quality. Known degradation of DNA due to long post-mortem preservation time, repeated or extensive thawing, and/or poor storage conditions. May also have been sequenced with poor success.
RNAlater vs DMSO
That's part of part name, and could be another part attribute eg https://github.com/ArctosDB/arctos/issues/1384#issuecomment-357044801 (which would be a bit more work but a much better representation of the material).
collectors to give us this information
That's exactly what NULL/no attribute means: "we do not have these data."
I'm still not sure "High DNA quality" and "-20 freezer" can coexist.
I agree with Joe, this has to do with tissue quality based on preservation time, that is what we might know about the specimen. We don't yet know that there is RNA or whole genomes contained in the tissue and labeling the samples like that will lead everyone to believe these tissues are known to contain good RNA or DNA or known to not contain those things. Also, adding preservation method into this scale is not appropriate. This should be about the quality of the tissue at the time of preservation. There should be a field within this attribute called method with choices of LN2, -80, -20, formalin, ethanol, DMSO, RNAlater, etc. I also agree with Dusty in that a grade of unknown or 0 is the same as no grade at all. So a scale from 1-5 or 1-6 is fine with me. Call it poor-excellent if you don't like numbers, that is fine with me too. My proposal, modified from Joe's: 6 - super --preserved within 10 minutes 5 - excellent --preserved within 1 hour 4- very good --preserved within 24 hours 3 - good --no obvious decomposition, multiple freeze thaw events 2 - fair --degraded from decomposition 1 - poor --decomposed or very rotton
quality of the tissue at the time of preservation.
There can be any number of these things - it's quality at the time of DETERMINED_DATE.
method
That doesn't exist...
UAM@ARCTOS> desc specimen_part_attribute
Name Null? Type
----------------------------------------------------------------- -------- --------------------------------------------
PART_ATTRIBUTE_ID NOT NULL NUMBER
COLLECTION_OBJECT_ID NOT NULL NUMBER
ATTRIBUTE_TYPE NOT NULL VARCHAR2(30)
ATTRIBUTE_VALUE NOT NULL VARCHAR2(4000)
ATTRIBUTE_UNITS VARCHAR2(30)
DETERMINED_DATE DATE
DETERMINED_BY_AGENT_ID NUMBER
ATTRIBUTE_REMARK VARCHAR2(4000)
...and as method those data would be essentially "1/rotten, because RNAlater..." which doesn't make much sense. "1/rotten because it stinks" or "1/rotten because DETERMINED_BY couldn't get mtDNA" might be useful, and adding method to this table wouldn't be too much of a problem if so.
LN2, -80, -20, formalin, ethanol, DMSO, RNAlater, etc. still sounds like another part attribute "preservation" or somesuch to me. (And you could have many of those as well - "ethanol" + "-80" would fit.)
I agree with Joe and Aren to simplify, except I would ideally do away with the timing qualifiers because we have to apply these codes to changes in quality that happen after the tissues are in the collection. An excellent tissue collected within 10 min in nitrogen is 6, but it goes to 5 if brought back and put in -80C, and goes to 4 after 1-2 loans, depending on how they were cut, and then to 1 after 20 years and two freezer failures. We should use a basic subjective assessment as below and then add details of preservation and storage in remarks or as additional attributes.
On Jan 12, 2018 5:31 PM, "Aren Gunderson" notifications@github.com wrote:
I agree with Joe, this has to do with tissue quality based on preservation time, that is what we might know about the specimen. We don't yet know that there is RNA or whole genomes contained in the tissue and labeling the samples like that will lead everyone to believe these tissues are known to contain good RNA or DNA or known to not contain those things. Also, adding preservation method into this scale is not appropriate. This should be about the quality of the tissue at the time of preservation. There should be a field within this attribute called method with choices of LN2, -80, -20, formalin, ethanol, DMSO, RNAlater, etc. I also agree with Dusty in that a grade of unknown or 0 is the same as no grade at all. So a scale from 1-5 or 1-6 is fine with me. Call it poor-excellent if you don't like numbers, that is fine with me too. My proposal, modified from Joe's: 6 - super --preserved within 10 minutes 5 - excellent --preserved within 1 hour 4- very good --preserved within 24 hours 3 - good --no obvious decomposition, multiple freeze thaw events 2 - fair --degraded from decomposition 1 - poor --decomposed or very rotton
— You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub https://github.com/ArctosDB/arctos/issues/1384#issuecomment-357392129, or mute the thread https://github.com/notifications/unsubscribe-auth/AOH0hDFRHYXwlcMSkS6uNGG6V6Ti69D6ks5tJ_lIgaJpZM4RTd6I .
I don't agree. Preserved within 10 minutes in 95% ethanol and not stored in LN2 does not have the same quality as preserved within 10 minutes in RNALater or in LN2 and stored in LN2. By your scale, those would be rated the same.
What would you do with a tissue that's 'preserved within 24 hours' in a -20 freezer, and then has just one freeze thaw event during processing? According to this, would you rate it as 'very good' ??? Same as something that's preserved within 2 hours (i.e., >1 hour) in LN2?
Carol brought up the case of a tissue where we might not know the preservation time/history, but someone sequenced it and was able to get good DNA out of it. What would you do with that - wouldn't fit into your criteria here.
Whatever we do, it needs to be interpreted the same way across collections AND people, so that there's consistency in how this is applied. Otherwise it loses meaning. I don't think we're there yet.
I think it has to do with fitness for use, which DOES depend on preservation and storage method in addition to preservation time.
in remarks
With attributes (or any controlled-data approach), "find {whatever} that has a condition of excellent and does not have any less-than-excellent condition determinations" is possible. With remarks, you can look for stuff where the arbitrary string someone typed matches the arbitrary string you feed to the search form....
I agree that remarks will be important to support these determinations, but "give me stuff from which I can probably get RNA" requires controlled vocabulary.
Your example would just have 4 determinations. Someone looking for tissues probably only cares about the latest one, but the historical determinations could be useful in analyzing your sampling techniques or predicting the lifespan of a sample or etc.
A tissue put in ethanol within 10 minutes is the same quality as a tissue put in LN2 after 10 minutes, the tissue is 10 minutes post mortem in both cases. The ethanol method allows the tissue to further degrade while the LN2 method keeps it's quality intact. Like Mariel said, the quality is going to change over time.
Almost everything we process in the lab will be a 3 if it were thawed for preparation and not rotten. Specimens processed in the field would be 4, 5, or 6.
I don't use part attributes so I thought there could be a field for preservation method. That should be another part attribute. If part attributes have a determined date then the history of preservation can recorded by adding a preservation method for every change in preservation condition. The tissue quality attribute can be edited whenever necessary to reflect to the tissues change in quality. This would be relatively easy at the time of sampling for loans or in some bulk-edit widget for all tissues in a freezer that failed.
To accommodate changes in tissue quality over time I think Mariel is right, the definitions should not contain a time component. That would leave this I suppose: 6 - super 5 - excellent 4- very good 3 - good --no obvious decomposition, multiple freeze thaw events 2 - fair --degraded from decomposition 1 - poor --decomposed or very rotton and maybe a scale of 1-4 would suffice.
It occurs to me that we are thinking of 'quality' in two fundamentally different ways. I was thinking of 'quality' in terms of fitness for use (from an informatics perspective re: data quality-see our VertNet paper). Others are thinking of quality in terms of putative degradation. They are of course related, and you can get at fitness for use by information on preservation/storage method (which is part of the part) + 'quality' (attribute). However, you cannot do the same types of studies with tissue in 95% etoh as you can with tissue in RNAlater/LN2 even if both are preserved shortly post-mortem.
I'm fine with defining quality in terms of putative degradation, but we need to make it clear that's the definition we're using. I also don't see any value in a 6-point scale. What is the difference between 'super' and 'excellent' and between 'excellent' and 'very good' ??? We need clear, practical definitions.
5 - excellent -- no obvious degradation, no freeze/thaw events 4 - very good -- no obvious degradation, one freeze/thaw event 3 - good --no obvious degradation, multiple freeze thaw events 2 - fair --degraded from decomposition 1 - poor --decomposed or very rotten
In this scenario, a tissue that's stored in LN2 and thawed once for subsampling would get the same #4 rating as a tissue taken from a carcass in good condition that's frozen at -20 and thawed once for preparation. Is that what we want? I'm ok with it, but we need consistency on how this would be applied.
I agree, which is why I think we need more than one attribute. We should also have attributes for fixation, preservation, and storage (LN2, -80C, -20C, dry, 95% EtOH, 70% EtOH, RNAlater, formalin-, etc), which would solve other problems with different part types. No one has to use these, but they would be available. Maybe we can come up with a way of concatenating the visual display to avoid overwhelming the parts table, and using a pop-up in data entry that allows selection of multiple part attribute values at time.
The problem with fitness for use vs degradation is that the former changes as technology changes, while the latter can reasonably be assessed objectively and subjectivity on standard criteria.
I proposed a category 6 for RNA, based on minimal to no degradation and immediate preservation at -196C and storage maintaining the cold chain at the same temperature. This has only become available to us in the last year when we moved to vapor phase nitrogen. Prior to that all our " Excellent" tissues were moved out of LN2 into -80 for storage. Of course, you can in theory get the same fitness for use from RNAlater at room temp, although I've heard of some issues with that. This is why we need a way to record all the collection time, fixation, preservation, and storage details so that fitness for use can be assessed based on the available technology at the time.
,
On Jan 13, 2018 1:20 PM, "Carla Cicero" notifications@github.com wrote:
It occurs to me that we are thinking of 'quality' in two fundamentally different ways. I was thinking of 'quality' in terms of fitness for use (from an informatics perspective re: data quality-see our VertNet paper). Others are thinking of quality in terms of putative degradation. They are of course related, and you can get at fitness for use by information on preservation/storage method (which is part of the part) + 'quality' (attribute). However, you cannot do the same types of studies with tissue in 95% etoh as you can with tissue in RNAlater/LN2 even if both are preserved shortly post-mortem.
I'm fine with defining quality in terms of putative degradation, but we need to make it clear that's the definition we're using. I also don't see any value in a 6-point scale. What is the difference between 'super' and 'excellent' and between 'excellent' and 'very good' ??? We need clear, practical definitions.
5 - excellent -- no obvious degradation, no freeze/thaw events 4 - very good -- no obvious degradation, one freeze/thaw event 3 - good --no obvious degradation, multiple freeze thaw events 2 - fair --degraded from decomposition 1 - poor --decomposed or very rotten
In this scenario, a tissue that's stored in LN2 and thawed once for subsampling would get the same #4 https://github.com/ArctosDB/arctos/issues/4 rating as a tissue taken from a carcass in good condition that's frozen at -20 and thawed once for preparation. Is that what we want? I'm ok with it, but we need consistency on how this would be applied.
— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/ArctosDB/arctos/issues/1384#issuecomment-357464551, or mute the thread https://github.com/notifications/unsubscribe-auth/AOH0hIzOv1CDL5d82qev3ZHu7noun-VGks5tKRAUgaJpZM4RTd6I .
d more than one attribute...also have attributes for fixation, preservation, and storage
They're free, you can have as many as you want!
(And FWIW I think you're correct - we keep looping back around to time-at-temperature; the longer this thread gets, the more that looks like one unavoidable component of whatever it is that we're trying to say.)
No one has to use these, but they would be available.
Just saying this again - ALL of this stuff is a way to better communicate what you have to users (including those filling loan etc.). You don't have to do anything, anywhere, and we can talk about tools (eg, https://github.com/ArctosDB/arctos/issues/1371) to make whatever you do want to do easier.
concatenating the visual display
Yea, perhaps we can summarize these data somehow, but I don't think it would be a full replacement for a table with details.
data entry that allows selection of multiple part attribute values at time
I think that's magic, but I'm willing to try whatever. And much of that's probably better done from the container end of things - eg, add an attribute for everything scanned into an LN2 tank in the field.
Here are the current options: 5 - excellent -- no obvious degradation, no freeze/thaw events 4 - very good -- no obvious degradation, one freeze/thaw event 3 - good --no obvious degradation, multiple freeze thaw events 2 - fair --degraded from decomposition 1 - poor --decomposed or very rotten
I suggest we stick with this scale, and then add preservation attributes and storage atributes that would distinguish the following
preservation:
liquid nitrogen, dry ice, -80C freezer, -20C freezer, refrigeration, dry, ethanol (95%, 80%, 70% etc),
isopropanol, RNAlater, EDTA buffer, Longmire's solution, etc. etc.
Storage:
liquid nitrogen, dry ice, -80C freezer, -20C freezer, refrigeration, room temp, dry, ethanol (95%, 80%,
70% etc), isopropanol, RNAlater, EDTA buffer, Longmire's solution, etc. etc.
That's easy to do, but we also have container environment available. Perhaps it's OK to have the generalization ("5 - excellent") and then go digging through containers when someone really needs the details, rather than trying to maintain a bunch of attributes for each of...
UAM@ARCTOS> select count(*) from specimen_part,ctspecimen_part_name where specimen_part.part_name=ctspecimen_part_name.part_name and is_tissue=1;
COUNT(*)
----------
5567955
... five and a half million individual parts?
Using container environment: globally shared - temperatures, etc. could be shared; we would need to let researchers see this info so that they can choose what types of samples to request Issue with part condition being visible to public for cultural collections; would need to make it possible to encumber
Clarification:
What is our status on this? I have a part recorded as excellent in part condition, that is obviously not anymore. We do not have any part attributes available yet for condition. This condition is a condition of a part, not of container environment, although I imagine that might have had something to do with the part condition at some point, but not recorded. Are we going to create part attributes for condition? If so, I propose we implement it now on 1-5 scale as above and test this out. I also think we need a way to integrate contatiner environment through a quick link from the parts menu in the specimen page. Can we make the barcodes a live link to the Find Container: Edit container page? Then we could quickly check environmental history, or add info.
Yes on 1-5 scale as above and test it out.
I'm all for a way to integrate container environment through a quick link from the parts menu in the specimen page. I like the idea of a live link with the barcode to an environmental history or add info.
Changing values isn't TOO bad, as long as no steps are lossy.
Changing the attribute name would be a pain - is " tissue sample quality " OK for everyone?
http://arctos-test.tacc.utexas.edu/guid/CHAS:Bird:17187 still exists, if anyone wants to see this (with bad values) in action.
as above
This:
5 - excellent -- no obvious degradation, no freeze/thaw events 4 - very good -- no obvious degradation, one freeze/thaw event 3 - good --no obvious degradation, multiple freeze thaw events 2 - fair --degraded from decomposition 1 - poor --decomposed or very rotten
where the value is "5 - excellent" and the definition is eg "no obvious degradation, no freeze/thaw events" - right?
I'll take a lack of objections by ~tomorrow as enthusiastic approval.
integrate
Barcodes are now clickable and open the container tree. Anything beyond that is deserving of a dedicated Issue.
http://arctos.database.museum/info/ctDocumentation.cfm?table=CTTISSUE_QUALITY http://arctos.database.museum/info/ctDocumentation.cfm?table=CTSPECPART_ATTRIBUTE_TYPE
Closing this; please open a new issue for any remaining problems, vocabulary adjustments, etc.
UAM mammals currently use:
5= in LN2 1 hr after death 4= frozen 24 hrs after death 3 = frozen and thawed 2 = somewhat rotten 1 = very rotten 0 = unknown/unchecked