plant2human workflow

main @ 76a6471 (latest)

• main @ 76a6471 (latest)
  
• main @ 1aa2763
  
• main @ b8c0b1d (earliest)
  

plant2human workflow ~ structural similarity vs. sequence similarity ~

This analysis workflow is centered on foldseek, which enables fast structural similarity searches and supports the discovery of understudied genes by comparing plants, which are distantly related species, with humans, for which there is a wealth of information. Based on the list of genes you are interested in, you can easily create a scatter plot of “structural similarity vs. sequence similarity” by retrieving structural data from the AlphaFold protein structure database.

Implementation Background

In recent years, with the AlphaFold protein structure database, it has become easier to obtain protein structure prediction data and perform structural similarity searches even for plant species such as rice. Against this background, searching for hits with “low sequence similarity and high structural similarity” for the gene groups being focused on has become possible. This approach may allow us to discover proteins that are conserved in distantly related species and to consider the characteristics of these proteins based on the wealth of knowledge we have about humans.

Analysis Environment

1. Using Dev Containers

Using the Dev Containers, you can create an analysis environment using a VScode extension. Please check the official website for details.

• Developing inside a Container
• Development Containers

2. Executing with cwltool

This analysis workflow is tested using cwltool version 3.1.20241112140730.

Example 1 ( Oryza sativa vs Homo sapiens)

Here, we will explain how to use the list of 10 rice genes as an example.

1. Creation of a TSV file of gene and UniProt ID correspondences

First, you need the following gene list tsv file. (Please set the column name as "From")

From
Os01g0187600
Os12g0129300
Os12g0159500
Os02g0609000
Os05g0468600
Os05g0352750
Os06g0140700
Os04g0391500
Os01g0795250
Os01g0859200

The following TSV file is required to execute the following workflow.

From	UniProt Accession
Os01g0187600	A0A0P0UZ77
Os12g0129300	A0A0P0Y6G7
Os12g0129300	B9GBP4
Os12g0159500	A0A0P0Y794
Os12g0159500	A0A8J8YJ44
Os12g0159500	B9GBZ8
...

To do this, you need to follow the CWL workflow command below. This yaml file is the parameter file for the workflow, for example.

cwltool --debug --outdir ./test/oryza_sativa_test ./Tools/01_uniprot_idmapping.cwl ./job/uniprot_idmapping_job_example_os.yml

In this execution, mmcif files are also retrieved. The execution results are output with the jupyter notebook.

2. Creating and Preparing Indexes

I'm sorry, but the main workflow does not currently include the creation of an index (both for foldseek index and BLAST index). Please perform the following processes in advance.

2-1. Creating a Foldseek Index

In this workflow, the target of the structural similarity search is specified as the AlphaFold database to perform comparisons across a broader range of species. Index creation using the foldseek databases command is through the following command.

Please select the database you want to use from Alphafold/UniProt, Alphafold/UniProt50-minimal, Alphafold/UniProt50, Alphafold/Proteome, Alphafold/Swiss-Prot. You can check the details of this database using the following command.

docker run --rm quay.io/biocontainers/foldseek:9.427df8a--pl5321hb365157_1 foldseek databases --help

For example, if you want to specify AlphaFold/Swiss-Prot as the index, you can do so with the following command,

# using docker container
docker run -u $(id -u):$(id -g) --rm -v $(pwd):/home -e HOME=/home --workdir /home quay.io/biocontainers/foldseek:9.427df8a--pl5321hb365157_1 foldseek databases Alphafold/Swiss-Prot swissprot tmp --threads 8

# making directory
mkdir ./index/index_swissprot

# moving index file
mv swissprot* ./index/index_swissprot/

Note: We have written a CWL file describing the above process, but we have confirmed an error and are correcting it.

2-2. Downloading a BLAST Index File

An index FASTA file must be downloaded to obtain the amino acid sequence using the blastdbcmd command from the UniProt database. Since it is a rice and human comparison, it can be downloaded as follows.

# Oryza sativa (all uniprot entries)
curl -o uniprotkb_39947_all.fasta.gz "https://rest.uniprot.org/uniprotkb/stream?compressed=true&format=fasta&query=%28organism_id%3A39947%29"

gzip -d uniprotkb_39947_all.fasta.gz

# Homo sapiens (all uniprot entries)
curl -o uniprotkb_9606_all.fasta.gz "https://rest.uniprot.org/uniprotkb/stream?compressed=true&format=fasta&query=%28organism_id%3A9606%29"

gzip -d uniprotkb_9606_all.fasta.gz

3. Execution of the Main Workflow

In this process, we perform a structural similarity search using the foldseek easy-search command and then perform a pairwise alignment of the amino acid sequences of the hit pairs using the needle and water commands. Finally, based on this information, we create a scatter plot and output a jupyter notebook as a report. Examples of commands are as follows.

cwltool --debug --outdir ./test/oryza_sativa_test ./Workflow/plant2human.cwl ./job/plant2human_job_example_os.yml

For example, you can visualize the results of structural similarity and global alignment, as shown below. In this case, the x-axis represents the global alignment similarity match (%), and the y-axis represents the LDDT score (an indicator of structural alignment).

https://github.com/yonesora56/plant2human/blob/main/image/rice_test_scatter_plot.png

The following scatter diagram can also be obtained from the test results of Zey Mays random 100 genes vs. Homo sapiens.

https://github.com/yonesora56/plant2human/blob/main/image/zey_mays_scatter_plot.png