Reduce dependencies (#45)

Dockerfile

@@ -16,11 +16,11 @@ RUN tar -xf ont-vbz-hdf-plugin-1.0.1-Linux-x86_64.tar.gz
 RUN conda env create -f environment.yml
 
 # Activate the environment
-RUN echo "conda activate lrst_py39" >> ~/.bashrc
+RUN echo "conda activate longreadsum" >> ~/.bashrc
 SHELL ["/bin/bash", "--login", "-c"]
 
 # Ensure the correct environment is being used
-RUN export PATH="/opt/conda/envs/lrst_py39/bin/python"
+RUN export PATH="/opt/conda/envs/longreadsum/bin/python"
 RUN which python
 
 # Build LongReadSum
@@ -30,4 +30,4 @@ RUN make
 ENV HDF5_PLUGIN_PATH="/longreadsum/lib/"
 
 # The code to run when container is started:
-ENTRYPOINT ["conda", "run", "--no-capture-output", "-n", "lrst_py39", "python", "/app/longreadsum"]
+ENTRYPOINT ["conda", "run", "--no-capture-output", "-n", "longreadsum", "python", "/app/longreadsum"]

environment.yml

@@ -1,4 +1,4 @@
-name: lrst_py39
+name: longreadsum
 channels:
  - bioconda
  - anaconda
@@ -9,12 +9,8 @@ dependencies:
  - hdf5
  - htslib
  - swig
- - matplotlib
- - plotly=4.14
+ - plotly
  - pytest
- - pip
- - pip:
- - kaleido
 
 # conda env create --file=environment.yml
-# conda activate lrst_py39
+# conda activate longreadsum

src/cli.py

@@ -85,9 +85,6 @@ def get_common_param(margs):
  try:
  if not os.path.isdir(output_dir):
  os.makedirs(output_dir)
- if not os.path.isdir(output_dir + '/' + getDefaultImageFolder()):
- os.makedirs(output_dir + '/' +
- getDefaultImageFolder())
 
  except OSError as e:
  this_error_str += "Cannot create folder for " + \
@@ -164,12 +161,11 @@ def fq_module(margs):
  if exit_code == 0:
  logging.info("QC generated.")
  logging.info("Generating HTML report...")
- plot_filepaths = create_base_quality_plots(fq_output, param_dict, "FASTQ: Basic statistics")
- for static in [True, False]:
- fq_html_gen = generate_html.ST_HTML_Generator(
- [["basic_st", "read_length_st", "read_length_hist", "base_st", "basic_info", "base_quality",
- "read_avg_base_quality"], "FASTQ QC", param_dict], plot_filepaths, static=static)
- fq_html_gen.generate_st_html()
+ plot_filepaths = plot(fq_output, param_dict, 'FASTQ')
+ fq_html_gen = generate_html.ST_HTML_Generator(
+ [["basic_st", "read_length_bar", "read_length_hist", "base_counts", "base_quality",
+ "read_avg_base_quality"], "FASTQ QC", param_dict], plot_filepaths, static=False)
+ fq_html_gen.generate_st_html()
 
  logging.info("Completed.")
  else:
@@ -208,15 +204,11 @@ def fa_module(margs):
  if exit_code == 0:
  logging.info("QC generated.")
  logging.info("Generating HTML report...")
- from src import fasta_plot
- plot_filepaths = fasta_plot.plot(fa_output, param_dict)
-
- # TODO: Unused 'static' variable results in redundant function call
- for static in [True, False]:
- fa_html_gen = generate_html.ST_HTML_Generator(
- [["basic_st", "read_length_st", "read_length_hist", "base_st", "basic_info"], "FASTA QC",
- param_dict], plot_filepaths, static=True)
- fa_html_gen.generate_st_html()
+ plot_filepaths = plot(fa_output, param_dict, 'FASTA')
+ fa_html_gen = generate_html.ST_HTML_Generator(
+ [["basic_st", "read_length_bar", "read_length_hist", "base_counts"], "FASTA QC",
+ param_dict], plot_filepaths, static=True)
+ fa_html_gen.generate_st_html()
 
  else:
  logging.error("QC did not generate.")
@@ -252,14 +244,13 @@ def bam_module(margs):
  if exit_code == 0:
  logging.info("QC generated.")
  logging.info("Generating HTML report...")
- from src import bam_plot
- plot_filepaths = bam_plot.plot(bam_output, param_dict)
+ plot_filepaths = plot(bam_output, param_dict, 'BAM')
 
- for static in [True, False]:
-  bam_html_gen = generate_html.ST_HTML_Generator(
-  [["basic_st", "map_st", "err_st", "read_length_st", "read_length_hist", "base_st", "basic_info",
-  "base_quality"], "BAM QC", param_dict], plot_filepaths, static=static)
-  bam_html_gen.generate_st_html()
+ # TODO: Add read average base quality plot (not currently generated by bam_plot.plot)
+ bam_html_gen = generate_html.ST_HTML_Generator(
+ [["basic_st", "read_alignments_bar", "base_alignments_bar", "read_length_bar", "read_length_hist", "base_counts", "basic_info",
+ "base_quality"], "BAM QC", param_dict], plot_filepaths, static=False)
+ bam_html_gen.generate_st_html()
 
  else:
  logging.error("QC did not generate.")
@@ -301,18 +292,17 @@ def seqtxt_module(margs):
  if exit_code == 0:
  logging.info("QC generated.")
  logging.info("Generating HTML report...")
- from src import seqtxt_plot
- plot_filepaths = seqtxt_plot.plot(seqtxt_output, param_dict)
- for static in [True, False]:
- if margs.seq == 0:
- f5_html_gen = generate_html.ST_HTML_Generator(
- [["basic_st", "read_length_st", "read_length_hist", "base_st", "basic_info"],
- "sequencing_summary.txt QC", param_dict], plot_filepaths, static=static)
- else:
- f5_html_gen = generate_html.ST_HTML_Generator(
- [["basic_st", "read_length_st", "read_length_hist", "basic_info"], "sequencing_summary.txt QC",
- param_dict], plot_filepaths, static=static)
- f5_html_gen.generate_st_html()
+ plot_filepaths = plot(seqtxt_output, param_dict, 'SeqTxt')
+
+ if margs.seq == 0:
+ seqtxt_html_gen = generate_html.ST_HTML_Generator(
+ [["basic_st", "read_length_bar", "read_length_hist", "base_counts", "base_quality", "basic_info"],
+ "sequencing_summary.txt QC", param_dict], plot_filepaths, static=False)
+ else:
+ seqtxt_html_gen = generate_html.ST_HTML_Generator(
+ [["basic_st", "read_length_bar", "read_length_hist", "basic_info"], "sequencing_summary.txt QC",
+ param_dict], plot_filepaths, static=False)
+ seqtxt_html_gen.generate_st_html()
  else:
  logging.error("QC did not generate.")
 
@@ -348,12 +338,12 @@ def fast5_module(margs):
  if exit_code == 0:
  logging.info("QC generated.")
  logging.info("Generating HTML report...")
- plot_filepaths = create_base_quality_plots(fast5_output, param_dict, "FAST5: Basic statistics")
- for static in [True, False]:
- fast5_html_obj = generate_html.ST_HTML_Generator(
-  [["basic_st", "read_length_st", "read_length_hist", "base_st", "basic_info", "base_quality",
-  "read_avg_base_quality"], "FAST5 QC", param_dict], plot_filepaths, static=static)
- fast5_html_obj.generate_st_html()
+ plot_filepaths = plot(fast5_output, param_dict, 'FAST5')
+ fast5_html_obj = generate_html.ST_HTML_Generator(
+ [["basic_st", "read_length_bar", "read_length_hist", "base_counts", "basic_info", "base_quality",
+ "read_avg_base_quality"], "FAST5 QC", param_dict], plot_filepaths, static=False)
+ fast5_html_obj.generate_st_html()
+
  else:
  logging.error("QC did not generate.")
 
@@ -390,13 +380,12 @@ def fast5_signal_module(margs):
  if exit_code == 0:
  logging.info("QC generated.")
  logging.info("Generating HTML report...")
- from src import fast5_signal_plot
- dynamic_plots, plot_filepaths = fast5_signal_plot.plot(fast5_output, param_dict)
-
- # Generate a dynamic HTML file
+ plot_filepaths = plot(fast5_output, param_dict, 'FAST5s')
  fast5_html_obj = generate_html.ST_HTML_Generator(
- [[], "FAST5 signal QC", param_dict], plot_filepaths, static=False)
- fast5_html_obj.generate_st_html(signal_plots=dynamic_plots)
+ [["basic_st", "read_length_bar", "read_length_hist", "base_counts", "basic_info", "base_quality",
+ "read_avg_base_quality", "ont_signal"], "FAST5 QC", param_dict], plot_filepaths, static=False)
+ fast5_html_obj.generate_st_html(signal_plots=True)
+
  else:
  logging.error("QC did not generate.")