• d4cb6f78 · delete unused cm_coords at updates ·

    v 1.5 beta, April 2021

    FEATURE CHANGES

    • New option --stats-opts="..." allows to pass options to the automatic run of statistics.py (when -s is used)
    • Removed support for 3'->5' Rfam hits, they are now completely ignored. They concern the opposite strand, which is unresolved in 3D.
    • Removed PyDCA, which is outdated and introduces dependencies conflicts. Code may be adapted later.
    • A new column in align_column table, index_small_ali, gives the index of the nucléotide in the "3d only" alignment.
    • 3D distance matrices are now computed only for match positions (match to the covariance model)

    BUG CORRECTIONS

    • Corrected a bug which skipped angle conversions from degrees (DSSR) to radians if nucleotides where renumbered.

    v 1.4 beta, March 2021

    Khodor Hannoush joins the development of RNANet.

    FEATURE CHANGES

    • SINA is now used only if you pass the option --sina, Infernal is used by default even for rRNAs.
    • A new option --cmalign-opts="..." allows to customize your cmalign runs, e.g. with --cyk. The default is no option.
    • RNANet makes use of PyDCA to compute DCA-related features on the alignments (descriptions to come in the Database.md)
    • statistics.py now fully supports the computation of 3D distance matrices, with average and standard deviation by RNA family
    • Now RNANet considers only the equivalence class representative structure by default. To consider all members of an equivalence class (like before), use the --redundant option.

    TECHNICAL CHANGES

    • cmalign is not run with --cyk anymore by default, and now requires huge amounts of RAM if launched with the default options.
    • Moving to a 60-core/128GB server for our internal runs.
  • a8cd46e8 · typo in 3D_only folder name ·

    v1.3-beta, January 2021

    The first uses of RNAnet by people from outside the development team happened last December. A few feedback allowed to identify issues and useful information to add.

    FEATURE CHANGES

    • Sequence alignments of the 3D structures mapped to a family are now provided.
    • Full alignments with Rfam sequences are not provided, but you can ask us for the files.
    • Two new fields in table 'family': ali_length and ali_filtered_length. They are the MSA lengths of the alignment with and without the Rfam sequences.
    • Gap replacement by consensus (--fill-gaps) has been removed. Now, the gap percentage and consensus are saved in the align_column table and the datapoints in CSV format, in separate columns. Consensus is one of ACGUN-, the gap being chosen if >75% of the sequences are gaps at this position. Otherwise, A/C/G/U is chosen if >50% of the non-gap positions are A/C/G/U. Otherwise, N is the consensus.

    TECHNICAL CHANGES

    • SQLite connections are now all in WAL mode by default (previously, only the writers used WAL mode, but this is useless)
    • Moved to Python3.9 for internal testing.
    • Latest version of BioPython is now supported (1.78)

    BUG CORRECTIONS

    • When an alignment file is updated in a newer run of RNANet, all the re_mappings are now re-computed for this family. Previously, the remappings were computed only for the newly added sequences, while the alignment actually changed even for chains added in past runs.
    • Changed the ownership and permissions of files produced by the Docker container. They were previously owned by root and the user could not get access to them.
    • Modified nucleotides were not always correctly transformed to N in the alignments (and nucleotide.nt_align_code fields). Now, the alignments and nt_align_code (and consensus) only contain "ACGUN-" chars. Now, 'N' means 'other' or 'any', while '-' means 'nothing' or 'unknown'.