Skip to content

Mixed Models - remedy for pseudoreplication?

5 messages · Jens Oldeland, ONKELINX, Thierry

Original
Jens Oldeland
# 
Dear List readers,

we are studying effects of chemical therapies on the proliferation of 
cancer cell lines.
My question aims at the practical application of a Mixed Effect Model 
(lme in R) of a particular analysis of a cell culture experiment on a 96 
well plate and the problem of pseudoreplication.

A 96 well plate looks like this: 
http://www.cellsignet.com/media/plates/96.jpg
and receives treatments in the columns 3-10 with 8 wells, i.e. the 
replications. We repeated the experiment on 6 days, with one 96 well 
plate per day.
We noticed a high intra-day variability.

Since all the cells of one day on which the treatments are applied come 
from the same flask and the media and pharmacologic agents, with which 
they were treated, are coming from the same tubes, the question arises 
if I have to count them as pseudoreplicates.
The analysis of the data using ?lme? works just fine, but I worry about 
the high number of degrees of freedom, which include the 
?pseudoreplicated? measurements.

An alternative would be to average the 8 replicates per day and use an 
ANOVA with an Error structure, which seems less elegant to me, since we 
are not including the high intra-day variability.

I would be very happy if anyone could give an adivce on how to deal with 
this dataset. I found different opinions in the literature, some saying 
that LMMs are a "remedy" for 96 well plates and some say that 96 well 
plates are per-se pseudoreplicated...

thanks in advance!
Jens
ONKELINX, Thierry
# 
Dear Jens,

You will need to provide more details on the design of the experiment.

- do you create 6 plates on day 0 and measure them on day 1, 2, ... or do you fill a new plate each day with the content of the flasks?
- how are the flasks distributed over the plates? Each well come for a different flask? Or all wells from the same column come from one flash?

We need all the details of the design to suggest a sensible model.

Best regards,

Thierry

ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature and Forest
team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
Kliniekstraat 25
1070 Anderlecht
Belgium
+ 32 2 525 02 51
+ 32 54 43 61 85
Thierry.Onkelinx at inbo.be
www.inbo.be

To call in the statistician after the experiment is done may be no more than asking him to perform a post-mortem examination: he may be able to say what the experiment died of.
~ Sir Ronald Aylmer Fisher

The plural of anecdote is not data.
~ Roger Brinner

The combination of some data and an aching desire for an answer does not ensure that a reasonable answer can be extracted from a given body of data.
~ John Tukey


-----Oorspronkelijk bericht-----
Van: r-sig-mixed-models-bounces at r-project.org [mailto:r-sig-mixed-models-bounces at r-project.org] Namens Jens Oldeland
Verzonden: dinsdag 30 juli 2013 12:21
Aan: r-sig-mixed-models at r-project.org
Onderwerp: [R-sig-ME] Mixed Models - remedy for pseudoreplication?

Dear List readers,

we are studying effects of chemical therapies on the proliferation of cancer cell lines.
My question aims at the practical application of a Mixed Effect Model (lme in R) of a particular analysis of a cell culture experiment on a 96 well plate and the problem of pseudoreplication.

A 96 well plate looks like this:
http://www.cellsignet.com/media/plates/96.jpg
and receives treatments in the columns 3-10 with 8 wells, i.e. the replications. We repeated the experiment on 6 days, with one 96 well plate per day.
We noticed a high intra-day variability.

Since all the cells of one day on which the treatments are applied come from the same flask and the media and pharmacologic agents, with which they were treated, are coming from the same tubes, the question arises if I have to count them as pseudoreplicates.
The analysis of the data using ?lme? works just fine, but I worry about the high number of degrees of freedom, which include the ?pseudoreplicated? measurements.

An alternative would be to average the 8 replicates per day and use an ANOVA with an Error structure, which seems less elegant to me, since we are not including the high intra-day variability.

I would be very happy if anyone could give an adivce on how to deal with this dataset. I found different opinions in the literature, some saying that LMMs are a "remedy" for 96 well plates and some say that 96 well plates are per-se pseudoreplicated...

thanks in advance!
Jens

--
+++++++++++++++++++++++++++++++++++++++++
Dr. Jens Oldeland

Post-Doc Researcher & Lecturer @ BEE
Managing Editor - Biodiversity & Ecology

Biodiversity, Ecology and Evolution of Plants (BEE) Biocentre Klein Flottbek and Botanical Garden University of Hamburg Ohnhorststr. 18
22609 Hamburg,
Germany

Tel:    0049-(0)40-42816-407
Fax:    0049-(0)40-42816-543
Mail:   jens.oldeland at uni-hamburg.de
         Oldeland at gmx.de
Skype:  jens.oldeland
http://www.biologie.uni-hamburg.de/bzf/fbda005/fbda005.htm
http://www.biodiversity-plants.de/biodivers_ecol/biodivers_ecol.php
+++++++++++++++++++++++++++++++++++++++++

_______________________________________________
R-sig-mixed-models at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
* * * * * * * * * * * * * D I S C L A I M E R * * * * * * * * * * * * *
Dit bericht en eventuele bijlagen geven enkel de visie van de schrijver weer en binden het INBO onder geen enkel beding, zolang dit bericht niet bevestigd is door een geldig ondertekend document.
The views expressed in this message and any annex are purely those of the writer and may not be regarded as stating an official position of INBO, as long as the message is not confirmed by a duly signed document.
Jens Oldeland
# 
Dear Thierry,

sorry for beeing imprecise, I provided clearer details below


- do you create 6 plates on day 0 and measure them on day 1, 2, ... or do you fill a new plate each day with the content of the flasks?
- how are the flasks distributed over the plates? Each well come for a different flask? Or all wells from the same column come from one flash?
So to say, there was one set of materials (flasks, cancer cells, tubes) that were mixed together on 6 different days, every day a new set uo was produced.

thank you already!
ONKELINX, Thierry
# 
Dear Jens,

I would suggest that you make a list of all elements in you design that can influence the result: e.g. treatment, tube, flask, day, plate, column, day:flask interaction, ...
- Eliminate the non-relevant ones (e.g. maybe column).
- Eliminate the ones that have only one level (flask?)
- Some will be confounding. If you have only one flask and do one plate each day, then you cannot separate the effect of day, plate and the interaction of day and flask. You have to choose one (e.g. day) and this will model the combined effect. If you want to separate those effects, then you will have to change your design.
- Make a model with the remain variables. lme(response  ~ treatment, random =  ~1|day) might be what you need.

Best regards,

ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature and Forest
team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
Kliniekstraat 25
1070 Anderlecht
Belgium
+ 32 2 525 02 51
+ 32 54 43 61 85
Thierry.Onkelinx at inbo.be
www.inbo.be

To call in the statistician after the experiment is done may be no more than asking him to perform a post-mortem examination: he may be able to say what the experiment died of.
~ Sir Ronald Aylmer Fisher

The plural of anecdote is not data.
~ Roger Brinner

The combination of some data and an aching desire for an answer does not ensure that a reasonable answer can be extracted from a given body of data.
~ John Tukey


-----Oorspronkelijk bericht-----
Van: Jens Oldeland [mailto:fbda005 at uni-hamburg.de]
Verzonden: dinsdag 30 juli 2013 16:14
Aan: ONKELINX, Thierry
CC: r-sig-mixed-models at r-project.org
Onderwerp: Re: [R-sig-ME] Mixed Models - remedy for pseudoreplication?

Dear Thierry,

sorry for beeing imprecise, I provided clearer details below


- do you create 6 plates on day 0 and measure them on day 1, 2, ... or do you fill a new plate each day with the content of the flasks?
- how are the flasks distributed over the plates? Each well come for a different flask? Or all wells from the same column come from one flash?
So to say, there was one set of materials (flasks, cancer cells, tubes) that were mixed together on 6 different days, every day a new set uo was produced.

thank you already!


--
+++++++++++++++++++++++++++++++++++++++++
Dr. Jens Oldeland

Post-Doc Researcher & Lecturer @ BEE
Managing Editor - Biodiversity & Ecology

Biodiversity, Ecology and Evolution of Plants (BEE) Biocentre Klein Flottbek and Botanical Garden University of Hamburg Ohnhorststr. 18
22609 Hamburg,
Germany

Tel:    0049-(0)40-42816-407
Fax:    0049-(0)40-42816-543
Mail:   jens.oldeland at uni-hamburg.de
         Oldeland at gmx.de
Skype:  jens.oldeland
http://www.biologie.uni-hamburg.de/bzf/fbda005/fbda005.htm
http://www.biodiversity-plants.de/biodivers_ecol/biodivers_ecol.php
+++++++++++++++++++++++++++++++++++++++++

* * * * * * * * * * * * * D I S C L A I M E R * * * * * * * * * * * * *
Dit bericht en eventuele bijlagen geven enkel de visie van de schrijver weer en binden het INBO onder geen enkel beding, zolang dit bericht niet bevestigd is door een geldig ondertekend document.
The views expressed in this message and any annex are purely those of the writer and may not be regarded as stating an official position of INBO, as long as the message is not confirmed by a duly signed document.
Jens Oldeland
# 
Dear Thierry,
In fact, this is exactly what we did. The problem we have that there 469 
degrees of freedom. By averaging there are only 6 dfs left.
Thus the question was "are we inflating our df's" due to 
pseudoreplication or are the single wells 'good' replications.

Due to the high intra-day variability, I was thinking a LME would be 
justified. but maybe I am wrong.

thank you Thierry
Jens


Am 30.07.2013 17:11, schrieb ONKELINX, Thierry: