Search
Question: MultiAssayExperiment: how to store data for multiple cancer lesions for each patients
0
gravatar for mario.zanfardino
21 days ago by
Naples (Italy)
mario.zanfardino90 wrote:

I have a data set of two experiments: A and B, stored in two summarizedExperiment of a MultiAssayExperiment. 

Experiments contains genomic data from cancer lesions. In some cases i have different data from two lesions of the same patient.

Where and how could i store these data? 

In the Summarized experiment: Each Assay of summarizedExperiment contains an observations. Each row of each assay contains genes. Each column of each assay contains Patients.

Thank you!

 

ADD COMMENTlink modified 20 days ago • written 21 days ago by mario.zanfardino90
2
gravatar for Levi Waldron
21 days ago by
Levi Waldron600
CUNY Graduate School of Public Health and Health Policy, New York, NY
Levi Waldron600 wrote:

Dear Mario,

I'm not sure I understand your question yet. It sounds like you have already stored these two SummarizedExperiment objects in a MultiAssayExperiment? Are you unable to access or manipulate the data in the way you want in the way you've constructed your MultiAssayExperiment, and if so, what do you want to do that you aren't able to do?

One comment of yours:

In some cases i have different data from two lesions of the same patient.

As a general principle, we designed the MultiAssayExperiment interface under the assumption that its colData should contain one row per biological unit. If you thought of your multiple lesions as simple technical replicates that you wanted to merge using the mean measurement, for example, then you could have one patient per row in the MultiAssayExperiment colData, associated to multiple columns of the SummarizedExperiments. You then have tools to identify and merge these replicates (?MultiAssayExperiment-helpers).

However it seems more likely that multiple lesions within a patient are distinct biological units, that you might want to model with random effects or something that explicitly models within-patient, across-patient within-group, and across-group variability. In this case you would want to represent each lesion as a row in the MultiAssayExperiment colData, with a "patient" column indicating which patient the lesion is from. When extracting using e.g. wideFormat(), you could maintain this "patient" column while keeping one row per lesion in the resulting wide-format DataFrame.

By the way, from what you've described, I see nothing unusual about your SummarizedExperiment and MultiAssayExperiment setup.

ADD COMMENTlink written 21 days ago by Levi Waldron600

Could you please provide an example of how to encode the following SummarizedExperiments in the MultiAssayExperiment colData.

SE1:

| biopsy | patient| lesion |
|--------|--------|--------|
|    1   |  HC    |    A   |

SE2:

| biopsy | patient| lesion |
|--------|--------|--------|
|    1   |  HC    |    A   |
|    1   |  HC    |    B   |
ADD REPLYlink modified 20 days ago • written 21 days ago by Lluís R370

This would be your MultiAssayExperiment colData (where HCA and HCB in the rownames would be replaced by unique identifiers if you repeat HCA and HCB as lesion identifiers in multiple biopsies.

biopsy lesion
HCA 1 HCA
HCB 1 HCB

And the MultiAssayExperiment sampleMap would look like this, although it can be auto-generated by the constructor in this simple example and in any case where the colnames of your SummarizedExperiments each map to a single row of the MultiAssayExperiment colData - you only have to specify the sampleMap yourself when that isn't the case:

assay primary colname
SE1 HCA HCA
SE2 HCA HCA
SE2 HCB HCB
ADD REPLYlink modified 20 days ago • written 20 days ago by Levi Waldron600
Seem to be having some Markdown issues. The colData table was intended to be: | | biopsy | lesion | |------|--------|--------| | HCA | 1 | HCA | | HCB | 1 | HCB |
ADD REPLYlink written 20 days ago by Levi Waldron600

Giving up on Markdown, here is the text colData:

 

|      | biopsy | lesion |
|------|--------|--------|
| HCA  |    1   |  HCA   |
| HCB  |    1   |  HCB   |

 

ADD REPLYlink written 20 days ago by Levi Waldron600

Ok, I'll try it next time. I wasn't sure it was possible to encode this. Mainly the biological unit can be interpreted differently: a biological unit is the type of lesion or the patient or the biopsy; depending on the question it can be either of them. Many thanks for resolving the issue.

PS: I didn't thought to use the code formatting... but I edited the question to make clearer the phenoData of each SE.

ADD REPLYlink modified 20 days ago • written 20 days ago by Lluís R370
1
gravatar for svlachavas
21 days ago by
svlachavas610
Greece/Athens/National Hellenic Research Foundation
svlachavas610 wrote:

Dear Mario, 

have you checked the tutorial of the MultiAssayExperiment R package: 

https://bioconductor.org/packages/release/bioc/vignettes/MultiAssayExperiment/inst/doc/QuickStartMultiAssay.html

as also the quick cheatsheet arguments and functions:

https://bioconductor.org/packages/release/bioc/vignettes/MultiAssayExperiment/inst/doc/MultiAssayExperiment_cheatsheet.pdf

 

Hope that helps,

Efstathios

ADD COMMENTlink modified 21 days ago • written 21 days ago by svlachavas610

The tutorials not answer my question!

ADD REPLYlink written 21 days ago by mario.zanfardino90
1
gravatar for mario.zanfardino
20 days ago by
Naples (Italy)
mario.zanfardino90 wrote:

Yes, multiple lesions within a patient are distinct biological units. My configuration In details:

RNAseq SE1:

Assay 1 (ex: count)

|            | Patient1 | Patient 2 |
|------------|----------|-----------|
| feature 1  |    5     |       7   |
| feature 2  |    10    |      66   |

 

Assay 2 (ex: log2 count)

|            | Patient1 | Patient 2 |
|------------|----------|-----------|
| feature 1  |    6     |       9   |
| feature 2  |    87    |      2    |

 

RNAseq SE2:

|            | Patient1 | Patient 2 |
|------------|----------|-----------|
| feature 1  |    5     |       7   |
| feature 2  |    10    |      66   |

 

MAE ColData:

|              |     PatientID    | vital_status |
|--------------|------------------|--------------|
| TCGA-A1-A0SB |    TCGA-A1-A0SB  |       7      |
| TCGA-A1-A0SB |    TCGA-A1-A0SB  |      66      |

 

This is my configuration. But! i have data for  RNAseq SE1 - assay 1 for two lesions (primary and metastatis of the same tissue)

How and where i could intert these data in my MAE?

 

ADD COMMENTlink written 20 days ago by mario.zanfardino90
2

colData would be something like:

|                |     PatientID    | vital_status | lesion  |
|----------------|------------------|--------------|---------|
| TCGA-A1-A0SB.1 |    TCGA-A1-A0SB  |       7      | primary |
| TCGA-A1-A0SB.2 |    TCGA-A1-A0SB  |       7      | mets    |
| TCGA-A1-A0SB   |    TCGA-A1-A0SB  |      66      | primary |

Note that since you're using different identifiers in the rownames of colData than in the colnames of the SummarizedExperiments, you'll have to provide a sampleMap to map these.

Also if you are working with TCGA data, note that the curatedTCGAData package and companion package TCGAutils provide pre-built MultiAssayExperiments and tools for things like splitting metastases into separate MultiAssayExperiment elements.

ADD REPLYlink modified 20 days ago • written 20 days ago by Levi Waldron600
0
gravatar for mario.zanfardino
20 days ago by
Naples (Italy)
mario.zanfardino90 wrote:

ADD COMMENTlink written 20 days ago by mario.zanfardino90
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 294 users visited in the last hour