Summarizing data quality/coverage across many runs

[jeffreybarrick]

Motivation: It would be very useful to have a script that can take many runs and create a dashboard for evaluating and comparing their quality/coverage.

It might:

• Generate a spreadsheet/table with summary statistics, like the number of reads/bases and the %mapping.
  – Show thumbnail coverage graphs across genomes
• Display the UN evidence concerning how much of the genome had enough coverage for calling mutations in each sample.
• etc.

Implementation: Most likely as Python/R scripts that generate HTML output. They can parse the summary.json files for statistics and use breseq BAM2COV to generate files to generate input files for graphing, for example.

[jeffreybarrick]

Here are some example summary files that can be used for testing:
https://barricklab.org/release/tmp/ADP1-summary.tgz

[jeffreybarrick]

HTML table as output.

Could eventually color some cells green/yellow/red to flag suspect files/samples.

In general, the output should have most of the same columns, but additional information, compared to the READ and REFERENCE tables generated for one breseq run. Example:

https://barricklab.org/twiki/pub/Lab/ToolsBacterialGenomeResequencing/REL8593A_output/summary.html

Columns to include in the REFERENCE TABLE:

• sample
• multiple lines within sample for each reference sequence
• length
• average coverage
• average fit coverage
• dispersion of fit coverage

Columns to include in the READ TABLE

• sample
• multiple lines within sample for each read file
• number of reads
• minimum-maximum [average] read length
• % mapping

[jeffreybarrick]

@ginnymortensen
Here is a newer set of breseq output that preserves all of the output folders compared to the one linked above. The output.json files are still the main place to pull information from.

https://barricklab.org/release/tmp/Ara-1-summary.tgz