The goal of this tutorial is to take raw NCBI data from some published research, convert the raw data into the AnnData format then add metadata to the object so that it can be used for further processing / analysis. Here we will look at the steps to obtain, understand, and manipulate the data in order for it to be properly processed.

The NCBI Database is a large repository for storing and accessing various types of biological data such as genome sequences, protein sequences, academic literature, etc. NCBI is a invaluable tool used for researchers and scientists, as a result it is important to know how to understand and utilise NCBI’s data resources.

Obtaining the Data

The first step is the obtain the data. For this tutorial, we will use data from the paper Song et al. 2022. The data for this research is stored in the Gene Expression Omnibus (GEO).

Hands-on: Choose Your Own Tutorial

This is a "Choose Your Own Tutorial" section, where you can select between multiple paths. Click one of the buttons below to select how you want to follow the tutorial

If you have experience finding and downloading data from GEO then you can use the Zenodo link to load the data directly into Galaxy, if you don't have experience with GEO then you can manually download, prepare, and load the data into Galaxy.

Manual Zenodo

We now have the raw gene expression data that we will process. However, we will need to manually add some metadata, which requires finding out some more information about our files.

Hands-on: Finding the metadata

1. Follow the link for the paper (https://pubmed.ncbi.nlm.nih.gov/35013146/), where you can see a link for accessing the full text. Accessing the full text takes us the following page https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8748675/

2. Downloading the paper and looking under Methods we can see that there are 2 biopsy samples for patients 1-3. For this tutorial we will be processing the first 3 patient’s data, therefore we will need to find more information about the biopsy samples.

Open image in new tab

1. On the NCBI page containing the paper we can scroll down to see a Supplementary Materials section. Under this section we should see various .xlsx files (Excel spreadsheets). These will contain metadata about the work done in the paper.

2. Looking through each file we can eventually find the spreadsheet 41467_2021_27322_MOESM2_ESM.xlsx which contains information about each patient including the biopsy sample IDs and tumor specimens. So we will want to download this in order to extract out the required information!

Open image in new tab

1. Download Supplementary Dataset 1 (41467_2021_27322_MOESM2_ESM.xlsx).

This one spreadsheet contains all the information we need to further understand our data.

Understanding the Data

If we look at that excel file 41467_2021_27322_MOESM2_ESM.xlsx, we can see multiple sheets.

Question

1. How many different patients are in this study?

Solution

1. Navigating to Clinical_info shows us that the data consists of 11 different patients, with some patients having multiple different samples of different tumor specimens.

For this tutorial, we are only going to process the first 3 patients. A summary of the relevant data can be seen below:

We can see that each patient has 2 different samples from different tumor locations. This is important information that needs to be added to our dataset so that we can separate it during analysis, if needed.

Unfortunately, the names of the files don’t exactly match the data in our excel sheet. This can make finding the right samples a bit difficult! A summary of which samples match which file can be found below:

• AUG_PB1A —> PA_AUG_PB_1A
• AUG_PB1B —> PA_AUG_PB_1B
• MAY_PB1A —> PA_PB1A
• MAY_PB1B —> PA_PB1B
• MAY_PB2A —> PA_PB2A
• MAY_PB2B —> PA_PB2B

You may also notice that we have multiple sample files with the same name suffixed with _Pool_X. These are replicates, where multiple samples are taken of the same tumor area in order to get a more comprehensive view of the gene data of the tumor. We will need to combine these replicates during processing.

These replicate files can be confirmed by looking under the Methods section of the paper where it states that one to four arrays were used per sample.

Open image in new tab

Creating the AnnData object

The next step is to convert all of the raw files into AnnData objects, this can be done one at a time or all at once by selecting multiple datasets when choosing the input and highlighting all the raw data files.

Tip: Tag your datasets

Before starting to process the data, it is generally a good idea to add tags to each sample in order to keep track of what data is being processed. Below is some of the imported data with added tags for the patient and sample id. We suggest you replicate this in your history.

Open image in new tab

Tip: Adding a tag

1. Click on the dataset to expand it
2. Click on Add Tags galaxy-tags
3. Add a tag starting with #
   • Tags starting with # will be automatically propagated to the outputs of tools using this dataset.
4. Press Enter
5. Check that the tag appears below the dataset name

Hands-on: Convert raw data to AnnData

1. Import AnnData and loom ( Galaxy version 0.7.5+galaxy1) with the following parameters:
   • “hd5 format to be created”: Anndata file
     • “Format for the annotated data matrix?”: Tabular, CSV, TSV
     • “Annotated data matrix”
       • param-files Multiple datasets: Select all imported files
     • “Does the first column store the row names?”: Yes
2. Inspect AnnData ( Galaxy version 0.7.5+galaxy1) with the following parameters:
   • param-file “Annotated data matrix”: select any one of the AnnData files
   • “What to inspect?”: Key-indexed observations annotation (obs)

Examine galaxy-eye the param-file “Inspect AnnData” output. You will find a list of genes - however, the obs (observations) layer in the AnnData object should store cell data. The var (variables) should store the genes data. We need to transpose all of our AnnData objects. We will speed up the process by selecting all of the AnnData objects and processing them at once.

Comment: Check your data

Whilst for this specific data, the object needed to be transposed, this won’t always be the case! The easiest way is to check the obs data of the AnnData object using the tool Inspect AnnData tool. The param-file obs file should have a column containing short nucleotide sequences (something like CGGAAGTGATAC). If that’s the case, then the data doesn’t need to be transposed, because those are the cell barcodes!

Hands-on: Transpose AnnData objects

1. Manipulate AnnData ( Galaxy version 0.7.5+galaxy1) with the following parameters:

   • “Annotated data matrix”
     • param-files Multiple datasets: Select all AnnData files
   • “Function to manipulate the object”: Transpose the data matrix, leaving observations and variables interchanged

Now, we have all the AnnData objects with the data in the correct orientation. We can combine them into a single AnnData object, which will make the data easier to work with. This combination operation will add an additional column called batch which tells us which AnnData object each bit of data came from. This will be useful for further processing!

Hands-on: Combine AnnData objects

1. Manipulate AnnData ( Galaxy version 0.7.5+galaxy1) with the following parameters:
   • param-file “Annotated data matrix”: Select first Manipulate AnnData (transpose) output
   • “Function to manipulate the object”: Concatenate along the observations axis
     • param-file “Annotated data matrix to add”: Select all other Manipulate AnnData (transpose) outputs
     • “Join method”: Intersection of variables
     • “Key to add the batch annotation to obs”: batch
     • “Separator to join the existing index names with the batch category”: -
2. Rename galaxy-pencil output Combined Object

Now we have all of our gene expression data stored in a single AnnData object! We just need to add our metadata!

Annotating the data

The next step is to annotate our data using the information gathered from the excel sheet earlier. We will do this by leveraging the batch column generated when the objects were combined for each individual patient/sample (indicated by the batch number). We will add in the relevant annotations. First, let’s extract the observation data so we can manipulate it.

Hands-on

1. Inspect AnnData ( Galaxy version 0.7.5+galaxy1) with the following parameters:
   • param-file “Annotated data matrix”: Combined Object
   • “What to inspect?”: Key-indexed observations annotation (obs)
2. Rename galaxy-pencil output Observation data

For each new column, we need to replace the relevant batch numbers with the appropriate metadata. After that, we can cut the modified column out so it is separated. We will repeat this process for each piece of metadata we want to add. All the relevant data can be found in the first table of this tutorial.

Let’s now add the replicate column which tells us which cells are part of pools of the same patient and tumor location.

Hands-on: Create replicate metadata

1. Replace Text ( Galaxy version 1.1.3) in a specific column with the following parameters:
   • param-file “File to process”: Observation data
   • “1: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 2|4|7|8
     • “Replace with”: poolA
   • + Insert Replacement
   • “2: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 3|5|9
     • “Replace with”: poolB
   • + Insert Replacement
   • “3: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 6
     • “Replace with”: poolC
   • + Insert Replacement
   • “4: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 0|1
     • “Replace with”: NA
   • + Insert Replacement
   • “5: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: batch
     • “Replace with”: replicate
2. Cut with the following parameters:
   • “Cut columns”: c2
   • “Delimited by”: Tab
   • param-file “From”: output of Replace Text
3. Rename galaxy-pencil output Replicate Metadata

Next we will add the metadata indicating which patient each row came from.

Hands-on: Create patient data

1. Replace Text ( Galaxy version 1.1.3) with the following parameters:
   • param-file “File to process”: Observation data
   • “1: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 0|1
     • “Replace with”: patient1
   • + Insert Replacement
   • “2: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 2|3|4|5|6
     • “Replace with”: patient2
   • + Insert Replacement
   • “3: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 7|8|9
     • “Replace with”: patient3
   • + Insert Replacement
   • “5: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: batch
     • “Replace with”: patient
2. Cut with the following parameters:
   • “Cut columns”: c2
   • “Delimited by”: Tab
   • param-file “From”: output of Replace Text
3. Rename galaxy-pencil output Patient Metadata

We will now add a column to indicate which sample each row came from using the sample ID’s described earlier.

Hands-on: Create sample ID metadata

1. Replace Text ( Galaxy version 1.1.3) with the following parameters:
   • param-file “File to process”: Observation data
   • “1: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 0
     • “Replace with”: AUG_PB1A
   • + Insert Replacement
   • “2: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 1$
     • “Replace with”: AUG_PB1B
   • + Insert Replacement
   • “3: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: (2$)|3
     • “Replace with”: MAY_PB1A
   • + Insert Replacement
   • “4: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 4|5|6
     • “Replace with”: MAY_PB1B
   • + Insert Replacement
   • “5: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 7
     • “Replace with”: MAY_PB2A
   • + Insert Replacement
   • “6: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 8|9
     • “Replace with”: MAY_PB2B
   • + Insert Replacement
   • “7: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: batch
     • “Replace with”: specimenID
2. Cut with the following parameters:
   • “Cut columns”: c2
   • “Delimited by”: Tab
   • param-file “From”: output of Replace Text
3. Rename galaxy-pencil output Specimen Metadata

Details: $ parameter

You may have noticed that some of the parameters in the previous tool used the $ symbol. This is due to how the Replace Text tool works. The tool will replace and update the data for every insert operation, meaning that lines that have already been updated could be updated again.

Since we are replacing our batch ID’s with identifiers that include numbers, we want to prevent them from being updated again. Appending the pattern with the $ symbol tells the tool to only replace the pattern if it is not followed by any other character (so the 1 in AUG_PB1A won’t be replaced as it is followed by an ‘A’).

Finally we will add the tumor column which indicates which tumor sample each row belongs to.

Hands-on: Create tumor metadata

1. Replace Text ( Galaxy version 1.1.3) with the following parameters:
   • param-file “File to process”: Observation data
   • “1: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 0
     • “Replace with”: left-mid
   • + Insert Replacement
   • “2: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 1|2|3|8|9
     • “Replace with”: right-mid
   • + Insert Replacement
   • “3: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 4|5|6
     • “Replace with”: right-apex
   • + Insert Replacement
   • “4: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: 7
     • “Replace with”: right-anterior
   • + Insert Replacement
   • “5: Replacement”
     • “in column”: Column: 2
     • “Find pattern”: batch
     • “Replace with”: tumorSpecimen
2. Cut with the following parameters:
   • “Cut columns”: c2
   • “Delimited by”: Tab
   • param-file “From”: output of Replace Text
3. Rename galaxy-pencil output Tumor Metadata

With all the individual metadata columns created, we can now combine them together to make a single tabular file containing our metadata. Since the Paste operation only allows us to combine two columns at once we will need to run the tool a few times to add all the columns together.

Hands-on: Combine metadata

1. Paste with the following parameters:
   • param-file “Paste”: Replicate Metadata
   • param-file “and”: Patient Metadata
   • “Delimit by”: Tab
2. Paste with the following parameters:
   • param-file “Paste”: Output of previous Paste
   • param-file “and”: Specimen Metadata
   • “Delimit by”: Tab
3. Paste with the following parameters:
   • param-file “Paste”: Output of previous Paste
   • param-file “and”: Tumor Metadata
   • “Delimit by”: Tab
4. Rename galaxy-pencil output Cell Metadata

After processing the Cell Metadata table should look like the following:

Open image in new tab

With the metadata table ready, we can add it to our original combined object!

Hands-on: Add metadata to AnnData object

1. Manipulate AnnData ( Galaxy version 0.7.5+galaxy1) with the following parameters:
   • param-file “Annotated data matrix”: Combined Object
   • “Function to manipulate the object”: Add new annotation(s) for observations of variables
     • “What to annotate?”: Observations (obs)
     • param-file “Table with new annotations”: Cell Metadata
2. Rename galaxy-pencil output Annotated Object

congratulations Congratulations! You have successfully interpreted the milieu of files in a published dataset, transformed the data into a usable format, and added all the relevant annotations to the AnnData object! All that’s left to do is to add some quality control metrics using automated tools!

Adding quality control metrics

With the manual annotations added, we need to do some further processing to add some statistical metadata about the genes. This is done automatically by running two different tools.

First, we will run the Scanpy FilterCells tool without actually filtering. This tool will add some metadata about the counts and numbers of expressed genes.

Warning: Scanpy FilterCells version issue

The Scanpy FilterCells ( Galaxy version 1.8.1+galaxy9) tool (used below) does not work in the latest version (1.8.1+galaxy93), switch to version 1.8.1+galaxy9 in order to run the tool without error.

Hands-on: Add initial metadata

1. Scanpy FilterCells ( Galaxy version 1.8.1+galaxy9) with the following parameters:
   • param-file “Input object in AnnData/Loom format”: Annotated Object
   • “Name of the column in anndata.var that contains gene name”: _index

The final tool, tool AnnData Operations, will add the rest of our metadata. This is mostly information about the mitochondrial genes present in the object, roughly counted by flagging genes that start with MT-.

Hands-on: Add final metadata

1. AnnData Operations ( Galaxy version 1.8.1+galaxy9) with the following parameters:
   • param-file “Input object in hdf5 AnnData format”: Output of Scanpy FilterCells
   • + Insert Flag genes that start with these names
   • “1: Parameters to select cells to keep”
     • “starts with”: MT-
     • “Var name”: mito

2. Rename galaxy-pencil output Final Object

3. Inspect AnnData ( Galaxy version 0.7.5+galaxy1) with the following parameters:
   • param-file “Annotated data matrix”: Final Object
   • “What to inspect?”: Key-indexed observations annotation (obs)

With that run we should be finished! Check the param-file obs file to see all the metadata that we’ve added in throughout the tutorial.

Warning: Check your final object!

The above tools will sometimes run without error but not produce the correct results - just because they run doesn’t mean all the parameters and inputs were set properly! When galaxy-eye inspecting the param-file obs file of the final AnnData object, you should see that all the columns contain data. If any of the columns are blank, then something has gone wrong!

congratulations Congratulations! Now your data is ready for further processing or analysis! You can compare your analysis results with this example history and redo the whole analysis with a single click using this workflow.

Key points

• Single cell data often requires additional processing and conversion before it is able to be analysed.

• Metadata and annotations can be added in both manually and automatically with various tools and resources.

Frequently Asked Questions

Useful literature

Further information, including links to documentation and original publications, regarding the tools, analysis techniques and the interpretation of results described in this tutorial can be found here.

References

1. Song, H., H. N. W. Weinstein, P. Allegakoen, M. H. Wadsworth, J. Xie et al., 2022 Single-cell analysis of human primary prostate cancer reveals the heterogeneity of tumor-associated epithelial cell states. Nature Communications 13: 10.1038/s41467-021-27322-4

Feedback

Did you use this material as an instructor? Feel free to give us feedback on how it went.
Did you use this material as a learner or student? Click the form below to leave feedback.

Citing this Tutorial

1. Morgan Howells, Converting NCBI Data to the AnnData Format (Galaxy Training Materials). https://training.galaxyproject.org/training-material/topics/single-cell/tutorials/scrna-ncbi-anndata/tutorial.html Online; accessed TODAY
2. Hiltemann, Saskia, Rasche, Helena et al., 2023 Galaxy Training: A Powerful Framework for Teaching! PLOS Computational Biology 10.1371/journal.pcbi.1010752
3. Batut et al., 2018 Community-Driven Data Analysis Training for Biology Cell Systems 10.1016/j.cels.2018.05.012

BibTeX

@misc{single-cell-scrna-ncbi-anndata,
author = "Morgan Howells",
	title = "Converting NCBI Data to the AnnData Format (Galaxy Training Materials)",
	year = "",
	month = "",
	day = ""
	url = "\url{https://training.galaxyproject.org/training-material/topics/single-cell/tutorials/scrna-ncbi-anndata/tutorial.html}",
	note = "[Online; accessed TODAY]"
}
@article{Hiltemann_2023,
	doi = {10.1371/journal.pcbi.1010752},
	url = {https://doi.org/10.1371%2Fjournal.pcbi.1010752},
	year = 2023,
	month = {jan},
	publisher = {Public Library of Science ({PLoS})},
	volume = {19},
	number = {1},
	pages = {e1010752},
	author = {Saskia Hiltemann and Helena Rasche and Simon Gladman and Hans-Rudolf Hotz and Delphine Larivi{\`{e}}re and Daniel Blankenberg and Pratik D. Jagtap and Thomas Wollmann and Anthony Bretaudeau and Nadia Gou{\'{e}} and Timothy J. Griffin and Coline Royaux and Yvan Le Bras and Subina Mehta and Anna Syme and Frederik Coppens and Bert Droesbeke and Nicola Soranzo and Wendi Bacon and Fotis Psomopoulos and Crist{\'{o}}bal Gallardo-Alba and John Davis and Melanie Christine Föll and Matthias Fahrner and Maria A. Doyle and Beatriz Serrano-Solano and Anne Claire Fouilloux and Peter van Heusden and Wolfgang Maier and Dave Clements and Florian Heyl and Björn Grüning and B{\'{e}}r{\'{e}}nice Batut and},
	editor = {Francis Ouellette},
	title = {Galaxy Training: A powerful framework for teaching!},
	journal = {PLoS Comput Biol} Computational Biology}
}

                   

Funding

These individuals or organisations provided funding support for the development of this resource

See Funder Profile

Go Further

Do you want to extend your knowledge? Follow one of our recommended follow-up trainings:

• Single Cell

Galaxy Administrators: Install the missing tools

You can use Ephemeris's shed-tools install command to install the tools used in this tutorial.

shed-tools install [-g GALAXY] [-a API_KEY] -t <(curl https://training.galaxyproject.org/training-material/api/topics/single-cell/tutorials/scrna-ncbi-anndata/tutorial.json | jq .admin_install_yaml -r)

Alternatively you can copy and paste the following YAML

---
install_tool_dependencies: true
install_repository_dependencies: true
install_resolver_dependencies: true
tools:
- name: text_processing
  owner: bgruening
  revisions: d698c222f354
  tool_panel_section_label: Text Manipulation
  tool_shed_url: https://toolshed.g2.bx.psu.edu/
- name: anndata_ops
  owner: ebi-gxa
  revisions: 890fb06a2893
  tool_panel_section_label: Single-cell
  tool_shed_url: https://toolshed.g2.bx.psu.edu/
- name: anndata_ops
  owner: ebi-gxa
  revisions: 825dfd66e3fb
  tool_panel_section_label: Single-cell
  tool_shed_url: https://toolshed.g2.bx.psu.edu/
- name: scanpy_filter_cells
  owner: ebi-gxa
  revisions: e9283529cba1
  tool_panel_section_label: Single-cell
  tool_shed_url: https://toolshed.g2.bx.psu.edu/
- name: unzip
  owner: imgteam
  revisions: 57f0914ddb7b
  tool_panel_section_label: Collection Operations
  tool_shed_url: https://toolshed.g2.bx.psu.edu/
- name: anndata_import
  owner: iuc
  revisions: fb9e030ffae4
  tool_panel_section_label: Single-cell
  tool_shed_url: https://toolshed.g2.bx.psu.edu/
- name: anndata_inspect
  owner: iuc
  revisions: ee98d611afc6
  tool_panel_section_label: Single-cell
  tool_shed_url: https://toolshed.g2.bx.psu.edu/
- name: anndata_manipulate
  owner: iuc
  revisions: 3d748954434b
  tool_panel_section_label: Single-cell
  tool_shed_url: https://toolshed.g2.bx.psu.edu/

Feedback