Jupyter is a program that can be started from the command line.

It starts a server (the hub) that will open in your browser (you've seen this). From there, you can start notebooks that run a kernel (e.g., Python session) which you can interact with through the browser.

# Starting jupyter from the command line
$ jupyter-notebook
    

Starting a Python notebook in jupyter

Select [New] and then [Python 3]

Getting started with notebooks

Code cells (Python), Markdown cells (rich text)

Getting started with notebooks

Code cells (Python), Markdown cells (rich text)

ipyrad CLI versus API

In the ipyrad CLI we use the '-n' flag

  # CLI (terminal): create a new Assembly named "test"
  $ ipyrad -n test
        

In the ipyrad API we create an Assembly object

  # API (Python session): create a new Assembly named "test"
  $ data = ip.Assembly("test")
        

Advantages to using the API

The API allows us to interactively set parameters, run assembly, etc.

  # create the Assembly object
  data = ip.Assembly("test")

  # show the params
  data.params
        

  0   assembly_name               test
  1   project_dir                 ~
  2   raw_fastq_path                           
  3   barcodes_path                      
  4   sorted_fastq_path
  5   assembly_method             denovo     
  6   reference_sequence                    
  7   datatype                    rad          
  8   restriction_overhang        ('TGCAG', '')  
  9   max_low_qual_bases          5          
  10  phred_Qscore_offset         33       
  ...
        

Advantages to using the API

Setting params can take advantage of tab-completion

  # set some parameters and try typing the first part and hitting [tab]
  data.params.raw_fastq_path = "/home/jovyan/ro-data/ipsimdata/gbs_example_R1_.fastq.gz"
  data.params.barcodes_path = "/home/jovyan/ro-data/ipsimdata/gbs_example_R1_.fastq.gz"
  data.params.datatype = "gbs"
  data.params
        

  0   assembly_name               test
  1   project_dir                 ~
  2   raw_fastq_path              /home/jovyan/ro-data/ipsimdata/gbs_example_R1_.fastq.gz  
  3   barcodes_path               /home/jovyan/ro-data/ipsimdata/gbs_barcodes.fastq.gz  
  4   sorted_fastq_path
  5   assembly_method             denovo     
  6   reference_sequence                    
  7   datatype                    gbs          
  8   restriction_overhang        ('TGCAG', '')  
  9   max_low_qual_bases          5          
  10  phred_Qscore_offset         33       
  ...
        

Branching: Multiple assemblies done easily

The seven steps of assembly

Branching: Multiple assemblies done easily

Start a different assembly from a previous checkpoint

Branching: Multiple assemblies done easily

Create a new branch from an existing assembly.

  # branching in the CLI
  $ ipyrad -p params-simdata.txt -b newdata
  

  loading Assembly: simdata
  from saved path: ~/Documents/ipyrad/tests/simdata.json
  creating a new branch called 'newdata' with 12 Samples
  writing new params file to params-newdata.txt
  

Branching to drop samples

Create a new branch from an existing assembly.

  # branching in the CLI
  $ ipyrad -p params-simdata.txt -b newdata - 1A_0 1B_0
  

  loading Assembly: simdata
  from saved path: ~/Documents/ipyrad/tests/simdata.json
  dropping 2 samples
  creating a new branch called 'newdata' with 10 Samples
  writing new params file to params-newdata.txt
  

Branching in the API

Create a new branch from an existing assembly.

  # branching in the API
  data1 = ip.Assembly("data1")
  data1.params.clust_threshold = 0.85

  # create a branch with different params
  data2 = data1.branch("data2")
  data2.params.clust_threshold = 0.90

  # run both assemblies through steps 1-7
  data1.run("1234567")
  data2.run("1234567")
  

API for analysis

Even if data was assembled w/ CLI, API can be useful afterwards.

  # load the assembly in the API
  data = ip.load_json("/home/jovyan/work/simdata.json")

  # show stats of the assembly
  data.stats

  # show output file paths of the assembly
  data.outfiles