NEWS.md

12 Nov 2020

• Better separation of the phases of the workflow
  • Raw data can be processed independently of further steps
  • Even without channel map or sample sequence
  • Or without any barcoding
  • We can detect the plugs without barcode
  • We can detect the barcode without identifying samples
• Control the workflow with new config options
  • run: attempt to run the whole typical workflow
  • init: only load the data
  • plugs: run only the plug detection
  • has_barcode: attempt to identify barcode plugs
  • has_samples_cycles: attempt to identify samples and cycles
  • samples_per_cycle: if run without a sequence file, you can provide here the number of samples in a cycle
• New parameters for the sample and cycle detection
  • min_end_cycle_barcodes: after how many consecutive barcode plugs we shall we start a new cycle
  • min_between_samples_barcodes: lowest number of barcode plugs separating two samples
  • min_plugs_in_sample: lowest number of plugs in a sample
• A new experimental method for barcode detection which aims to deal with the irregular, noisy data. This method doesn't work perfectly, requires fine-tuning of the parameters and normally we shouldn't use it as a decent quality experiment can be processed with the conventional methods. New parameters in config:
  • barcoding_method: either simple or adaptive
  • barcoding_param: dictionary with parameters for the barcoding method The simple method is the old behaviour and has only one parameter:
  • times: the blue channel should be at least this times higher than the control channel for barcode plugs With the adaptive method you can choose an alternative method and also scan a parameter space for this method.
• From now we save the results (plots) simply in a designated directory, by default results, without creating subdirectories for each run adding timestamp to the directory names. If you need subdirectories and timestamps use the following options:
  • result_subdirs
  • timestamp_result_subdirs
• A new option to make plugy silent, i.e. not to flood the console with log messages. We always have the log messages in the log file.
  • log_to_stdout: we keep the old behaviour, if you want to make it silent set this option to False
• No longer need to manually start the logging, it happens automatically
• No need to define paths with pathlib, you can simply provide them as strings
• It is possible to apply an independent second color scale to one compound on the heatmap; this is necessary because the positive control sometimes gives much higher signal than the drug compounds. This is controlled by the new config parameter heatmap_second_scale, by default it is set to pos_ctrl.
• We created a way to merge multiple experiments. Briefly, the first experiment will be the main one and all the subsequent ones should be merged (appended) to this one. It means first you need to create the second, third, etc experiments and run them only until the sample identification. Then you create the first experiment, and with the new config parameter append you provide all the subsequent experiments. It sounds complicated, but we will show an example.