sklearn_pca

A Python script to perform PCA and generate bidimensional plots between the three first components.

A Docker image is available at singgroup/sklearn_pca.

Motivation and example

Let's start with a TSV file like this (test_data/data1.tsv):

sample  f1      f2      f3
s1      1        2       33
s2      4        5       6
s3      7        8       9
s4      10      11      12

And a CSV file with associated metadata (test_data/metadata.csv) fr the samples:

sample,annotation1,annotation2
s1,A,C
s2,B,C
s3,A,D
s4,B,D

We would like to perform PCA on this table and colour samples by column annotation1 (this is --group) and give them different shapes based on the annotation2 column (this is --shape):

docker run --rm -v $(pwd):$(pwd) -w $(pwd) \
    singgroup/sklearn_pca pca \
        test_data/data1.tsv \
        --metadata test_data/metadata.csv \
        --group annotation1 \
        --shape annotation2 \
        --output_dir=test_data/test_docker_data1

In case your input data must be transposed, as with the test_data/data2.tsv file, use also --transpose:

docker run --rm -v $(pwd):$(pwd) -w $(pwd) \
    singgroup/sklearn_pca pca \
        test_data/data2.tsv \
        --transpose \
        --metadata test_data/metadata.csv \
        --group annotation1 \
        --shape annotation2 \
        --output_dir=test_data/test_docker_data2