Skip to content

Latest commit

 

History

History
57 lines (50 loc) · 3.5 KB

ChangeLog.md

File metadata and controls

57 lines (50 loc) · 3.5 KB
Raw

Maestro changes

  • Version 0.9 October 19, 2020

    • Improved the buffering of the transposed bitmatrix in build_db() to significantly reduce the number write operations (and hopefully fix the issues with Lustre/FSX stalling/hanging ...).

  • Version 0.8 September 13, 2020

    • Provide for separate scratch paths: --scratch.bloom and --scratch.database
    • Fixed template specialization-related compiler errors on g++
    • Added the option to --skip user-specified SRA run accessions. This is useful when trying to deal with accessions that cannot be downloaded and/or parsed by the SRA toolkit (i.e. ERR1197571).
      • At some point, we will need to add an "include" option to be able process previously skipped accessions (should they get fixed and become possible to include again).
    • Fixed bug in final database creation step (when all SRA records have been processed)

  • Version 0.7 August 29, 2020

    • Track worker memory usage to check for memory leaks
    • Even when streaming SRA records, remove *.cache and other SRA record-specific files.

  • Version 0.6 August 15, 2020

    • Handle aligned colorspace SRA records as a special case. These records cannot be read using the strategy of loading primary alignments followed by unaligned reads.
    • Use the SRA metadata to extract the number of bases and scale the size of the counting Bloom filters (to obtain the desired false positive rate).

  • Version 0.5 August 5, 2020

    • Added code to track the:
      • "deflation"; the number of Bloom filter bytes/number of sequence bytes
      • rate of SRA processing in kmers/second and bp/sec
      • ratio of kmers/bp for each Bloom filter -- what fraction of sequence is unique?
    • Modified the process_event() loop to automatically retry failed download and Bloom filter attempts (up to the number of allowed retries).
    • Added an option to force a retry of (STATUS_BLOOM_FAILURE, i.e. "hard") failed Bloom filters. Use "--retry.bloom" to retry all failed Bloom filters.
    • Added a new command line option, "--delay ", to ensure that download and/or streaming requests do not occur more frequently than once per the number of specified seconds.
    • Added additional reporting to the make_bloom_filter() function to track the progress reading through the SRA record. This will hopefully inform on the utility of restarting streaming failures during Bloom filter construction.
    • Catch SRA NGS error messages to help diagnose connectivity issues

  • Version 0.4 July 30, 2020

    • Modified the restore_bloom() in maestro_main.cpp to restore SRA accessions that are labeled as STATUS_DATABASE_FAIL (in addition to STATUS_BLOOM_SUCCESS). This will recover from previous database creation failures.
    • Moved some file I/O code to a new source file (file_io.cpp) from the maestro_main.cpp file.
    • Created a new help program, "manual_db", to update the accession in a status file with the accessions in a database file that is being manually copied to S3. This is needed when we get an upload failure (due to "aws s3 mv/cp" failing to upload a database file -- most likely due to Lustre FSX being overtaxed during the creation of a large databaes file.)

SRA inventory changes

  • Version 0.7 September 24, 2020
    • Added the option to specify an optional list of SRA run accessions to include.

CALDERA changes

  • Version 0.4c December 10, 2020
    • Changed program name from bigsi++ to caldera.
  • Version 0.4b September 25, 2020
    • Fixed supurious output when the input query is too small for a single kmer.