PICRUSt2 v2.4.1

Quick release to fix bug that prevented showing shuffle_predictions.py help documentation.

PICRUSt2 v2.4.0

Major changes:

• Added SEPP functionality, which can now be used as a low-memory alternative to EPA-ng for placing sequences

• Added in norm_taxon_function_contrib column which is the same as the column ContributionPercentOfSample in the PICRUSt1 metagenome contributions output (which is also present in the converted legacy file now too)

• Added a new script (shuffle_predictions.py), which can be used to quickly get shuffled trait predictions for comparing with actual predictions

Minor changes:

• Added edge_exponent as an HSP option (which is particularly useful sometimes when running SEPP)

• Removed white space from end of FASTA headers to avoid common error

• Allow for cases where no pathways are output

• Automatically check if any tips match trait table when quotes removed and remove if so

• Check if only very few rows of trait table found in tree

• Throw error if all NAs for any trait and warning if any NA in prediction table

• Default reference FASTA is now unzipped to make it compatible with SEPP

• Clarified message about sequences below NSTI cut-off to make it less confusing

• Fixed recurrent typo of "Enzyme Classification" to be "Enzyme Commission"

• Suppress printing log now when running MinPath (it was dominating the log output previously and making it hard to see other, more pertinent, messages)

• Increased version numbers used for dependencies and made minor syntax changes to get them working, esp. pandas. This was done so that the tool would be compatible with the latest QIIME 2 version.

PICRUSt2 v2.3.0-b

Major changes

• Now screen out study sequences that align poorly to reference sequences (taken to be 80% of original length by default). These sequences aren't placed at all and are listed in standard error.

• Fixed bug where a mix of integers and floats were all converted to floats when used as sequence ids.

Minor changes

• Improved how stderr and stdout of subprocess commands is captured when running picrust2_pipeline.py.

• Will output range of study sequence lengths when --verbose option set.

• Bumped r-castor version from 1.4.1 to 1.5.0

• Bumped epa-ng version from 0.3.5 to 0.3.6

• Bumped gappa version from 0.4.0 to 0.5.1

• Set required hmmer version range to be >=3.1.b2,<=3.2.1 rather than just 3.2.1 (Confirmed that study sequence alignments are identical with these three hmmalign versions: 3.1b2, 3.2, 3.2.1)

• Will now check for a column called "taxonomy" and throw a warning if it exists.

• Added more verbose options and changed option to be --verbose rather than --print_cmds for clarity in individual scripts.

• Throw error if space found in input FASTA filepath (causes problems for downstream placement tools)

• Create pipeline output directory after running basic sanity checks (like checking that all files exist)

• Set upperbound of 0.25.0 for pandas version for now to avoid massive spike in system time usage with newest versions

PICRUSt2 v2.2.0-b

Major:

• Long-format ("contributional") stratified output is now the default output stratified table, which greatly reduces memory usage. The wide-format table can still be output when the "--wide_table" option is set.

• Added option to skip copy number normalization to metagenome_pipeline.py and picrust2_pipeline.py along with new unittests.

• Fixed bug that caused error when running --per_seq_contrib with picrust2_pipeline.py.

• Bumped r-castor to 1.3.6 -> 1.4.1 and gappa from 0.2.0 -> 0.4.0

• Convert ASV ids to be strings to avoid cases where they are treated as floats.

• Changed output format of "metagenome contributions" filetype. Now both relative and raw ASV abundances are output and column names are simplified.

Minor:

• Removed exact pinning for most conda dependencies to fit with bioconda distribution best-practices.

• Added pre-print citation in setup.py

• Added garbage collector (i.e. gc()) calls to castor R script to help reduce memory usage.

• Now output an additional unstrat pathway abundance table based on regrouping per-seq-contrib stratified output when --per_sequence_contrib used.

• Added ability for convert_table.py to convert to legacy contributional format.

• Made option usage for add_descriptions.py clearer.

• Moved all MetaCyc pathway descriptions to single file.

• Gzipped all testfiles.

• Added test for --per-seq-contrib when running full pipeline.

• Fixed bug where samples with no ASVs / predictions were blank in the weighted NSTI output (now output as 0).

PICRUSt2 v2.1.4-b

Major changes:

• Added metagenome_contrib option to pipeline script (i.e. a similar breakdown of taxonomic contributors to functions as output in PICRUSt1).
• Changed default output files to be gzipped, which greatly saves on disk usage.
• Bumped to castor version 1.3.6.

Thanks to @efratmuller for suggesting the option to remove intermediate files and for an implementation in R for getting metagenome contributions to compare against.

Minor changes:

• Added flag to picrust2_pipeline.py to optionally remove intermediate files.
• Changed pd.read_table calls to pd.read_csv to avoid future warning.
• Removed unused tutorial folders.
• Removed unnecessary future imports.
• Changed pipeline timing to only be printed if verbose option set.
• Minor typo and help documentation changes.

PICRUSt2 v2.1.3-b

Major changes:

• Added step to parse jplace file so that EPA-NG produces more consistent output
• Added IMG phenotypes as trait table option
• Swapped in different way to convert biom to pandas dataframe to avoid numpy warning
• Updated requirements to match latest QIIME2 version (2019.4)
• Added option to allow reaction function to be specified in picrust2_pipeline.py

Minor changes:

• Added warning when non-default reference files used with default pathway mapfile
• Changed --threads options to be --processes across all scripts
• Renamed fungi reference folders to match filenames

PICRUSt2 v2.1.2-b

• Updated required package versions to match QIIME2 2019-0.1 environment

• Now output clear error when no tips in tree have known traits (i.e. mismatched reference files specified)

• Added support for mothur shared files.

• Added sort=True to all instances of pd.concat to avoid that associated warning.

• Clarified descriptions of several scripts and added additional usage examples.

PICRUSt2 v2.1.1-b

Misc changes:

• Bumped versions of GAPPA and EPA-NG to most recent

• Added jinja as conda environment requirement to avoid h5py warning

Added numerous experimental reference files:

• KEGG modules and pathways mapfiles and descriptions

• IMG phenotype counts

• Fungi EC and 18S/ITS copy number abundance tables

Bug fixes:

• Fixed bug when using unstructured mapping file that resulted in error (due to changes in last release)

• Fixed bug in pathway_pipeline.py that caused error when using --per-seq-contrib option when ASV names are numeric.

• Ensure that permissions on temporary MinPath directory are correct (which was causing problems on certain clusters - see #44

• Also now create intermediate output dir outside of loop in pathway_pipeline.py, which was causing unnecessary overhead before

• Removed mention of "descriptions" option from picrust2 pipeline script

PICRUSt2 v2.1.0-b

Updates in this release are overviewed here: #58

PICRUSt2 v2.0.4-b

• Added GAPPA and EPA-NG to requirements

• Removed placement tools repository

• Fixed test scripts to check for test files relative to scripts

• Changed location of default_files directory