weixin_39683241
weixin_39683241
2021-01-02 16:11

Trinity

  • [x] I have read the guidelines for bioconda recipes.
  • [ ] This PR adds a new recipe.
  • [x] AFAIK, this recipe is directly relevant to the biological sciences (otherwise, please submit to the more general purpose conda-forge channel).
  • [x] This PR updates an existing recipe.
  • [ ] This PR does something else (explain below).

Trinity bioconda update for version 2.8.1

该提问来源于开源项目:bioconda/bioconda-recipes

  • 点赞
  • 写回答
  • 关注问题
  • 收藏
  • 复制链接分享
  • 邀请回答

25条回答

  • weixin_39683241 weixin_39683241 4月前

    Is there a bioconda trinity channel or somewhere to discuss bioconda trinity-related things w/ any and all stakeholders?

    点赞 评论 复制链接分享
  • weixin_39861823 weixin_39861823 4月前

    There is no Trinity-specific channel, but the official Bioconda Gitter channel is https://gitter.im/bioconda/Lobby .

    点赞 评论 复制链接分享
  • weixin_39942191 weixin_39942191 4月前

    we love to see upstream dev taking over recipes, that is super cool. But there are a few problems with your update that makes it not usable for many people.

    The reason it was on the blacklist is, that the perl dependencies has been broken. The last weeks the community fixed many of those to a degree that I'm confident to update the old recipes and take it down from the blacklist. I do not think that removing tests is healthy for any recipes just to make it work.

    As mentioned we do like to include even optional dependencies to make pkgs work from the beginning. If this is not in your interest we could create two recipes, one trinity and a second trinity-scripts. Or something along these lines.

    Your TRINITY_HOME is currently not working as you need to set it in an additional activate.sh script (e.g. https://github.com/bioconda/bioconda-recipes/blob/master/recipes/rnasnp/scripts/activate.sh) but this more or less considered a hack. Maybe I don't get how TRINITY_HOME is supposed to work in your recipe :( A test that tests your behavior would be nice as well.

    
    bag:~/projects/code/bioconda-recipes$ conda activate trinity
    (trinity) bag:~/projects/code/bioconda-recipes$ env | grep TRINIT
    (trinity) bag:~/projects/code/bioconda-recipes$ env | grep HOME
    JAVA_HOME=/home/bag/miniconda3/envs/trinity
    HOME=/home/bag
    JAVA_HOME_CONDA_BACKUP=
    

    In addition to the conda package we generate containers (Docker, Singularity, rkt) for the community. All of those do not have the TRINITY_HOME set by default, even if you include the hack mentioned above.

    Disclaimer - I am relatively new to bioconda, and so still learning the ropes to a large extent.

    Thats completely fine and why we raising the issues here and want to help to make Trinitiy rock in Conda!

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    All sounds good. If anyone wants to help me get this latest release in while maintaining some of the stuff that I had tossed earlier, I'm happy to take a look or do whatever's needed here, and work with you all on future updates.

    best,

    ~b

    On Mon, Aug 20, 2018 at 4:58 PM Björn Grüning wrote:

    https://github.com/brianjohnhaas we love to see upstream dev taking over recipes, that is super cool. But there are a few problems with your update that makes it not usable for many people.

    The reason it was on the blacklist is, that the perl dependencies has been broken. The last weeks the community fixed many of those to a degree that I'm confident to update the old recipes and take it down from the blacklist. I do not think that removing tests is healthy for any recipes just to make it work.

    As https://github.com/yesimon mentioned we do like to include even optional dependencies to make pkgs work from the beginning. If this is not in your interest we could create two recipes, one trinity and a second trinity-scripts. Or something along these lines.

    Your TRINITY_HOME is currently not working as you need to set it in an additional activate.sh script (e.g. https://github.com/bioconda/bioconda-recipes/blob/master/recipes/rnasnp/scripts/activate.sh) but this more or less considered a hack. Maybe I don't get how TRINITY_HOME is supposed to work in your recipe :( A test that tests your behavior would be nice as well.

    bag:~/projects/code/bioconda-recipes$ conda activate trinity (trinity) bag:~/projects/code/bioconda-recipes$ env | grep TRINIT (trinity) bag:~/projects/code/bioconda-recipes$ env | grep HOME JAVA_HOME=/home/bag/miniconda3/envs/trinity HOME=/home/bag JAVA_HOME_CONDA_BACKUP=

    In addition to the conda package we generate containers (Docker, Singularity, rkt) for the community. All of those do not have the TRINITY_HOME set by default, even if you include the hack mentioned above.

    Disclaimer - I am relatively new to bioconda, and so still learning the ropes to a large extent.

    Thats completely fine and why we raising the issues here and want to help to make Trinitiy rock in Conda!

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bioconda/bioconda-recipes/pull/10425#issuecomment-414460932, or mute the thread https://github.com/notifications/unsubscribe-auth/AHMVXzYmm-4vAzKLxqkx5F-ddziGMs8Sks5uSyLhgaJpZM4V8-DQ .

    --

    Brian J. Haas The Broad Institute http://broadinstitute.org/~bhaas http://broad.mit.edu/~bhaas

    点赞 评论 复制链接分享
  • weixin_39559369 weixin_39559369 4月前

    Hi, I'd too much prefer to keep the scripts in $PREFIX/bin rather than relying on the TRINITY_HOME variable, even if we do it for a subset of the scripts. I think in the long run it would be better to find a trick to allow the scripts to run without the need for this TRINITY_HOME variable So, do we just add back the symlinks in the trinity recipe, or do you prefer to split it in a trinity-scripts recipe? I can help for both (and vote for the first one for compatibility with existing stuff) I too think rebuilding the older recipes should work, or be easy to fix now that many perl packages have been rebuilt, I can make a PR to add them back

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    Hi Anthony

    Please feel free to update the recent trinity recipe.

    Many thx

    -Brian (by iPhone)

    On Aug 21, 2018, at 6:21 AM, Anthony Bretaudeau wrote:

    Hi, I'd too much prefer to keep the scripts in $PREFIX/bin rather than relying on the TRINITY_HOME variable, even if we do it for a subset of the scripts. I think in the long run it would be better to find a trick to allow the scripts to run without the need for this TRINITY_HOME variable So, do we just add back the symlinks in the trinity recipe, or do you prefer to split it in a trinity-scripts recipe? I can help for both (and vote for the first one for compatibility with existing stuff) I too think rebuilding the older recipes should work, or be easy to fix now that many perl packages have been rebuilt, I can make a PR to add them back

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    I'm curious to test it out once the others are versions (subdirectories) are added back. There's one early version 'date.2011_11_26' that was bundled into one of the version subdirectories in the earlier version, but it should have been included as a separate subdirectory along with the others.

    I have the older version of it here: https://github.com/trinityrnaseq/trinityrnaseq/tree/master/bioconda_recipe/trinity

    in case that's helpful. Just let me know what you'd like me to do, if anything here.

    best,

    ~brian

    点赞 评论 复制链接分享
  • weixin_39861823 weixin_39861823 4月前

    I too think rebuilding the older recipes should work, or be easy to fix now that many perl packages have been rebuilt, I can make a PR to add them back.

    I'm not particularly a fan of the old recipes as subdirectories unless there is a use case, we always have the git history for looking at changes in the recipe.

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    I'd prefer to not have the subdirectories in there, but if there's a need for them by others, happy to continue to keep them there. We should have a roadmap or plan for when we can safely remove them, though, so they're not there for perpetuity.

    ~b

    On Tue, Aug 21, 2018 at 9:27 AM Nicola Soranzo wrote:

    I too think rebuilding the older recipes should work, or be easy to fix now that many perl packages have been rebuilt, I can make a PR to add them back.

    I'm not particularly a fan of the old recipes as subdirectories unless there is a use case, we always have the git history for looking at changes in the recipe.

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bioconda/bioconda-recipes/pull/10425#issuecomment-414673337, or mute the thread https://github.com/notifications/unsubscribe-auth/AHMVX33BA1f0S_gCihAf4wY7wZDaKK0lks5uTArbgaJpZM4V8-DQ .

    --

    Brian J. Haas The Broad Institute http://broadinstitute.org/~bhaas http://broad.mit.edu/~bhaas

    点赞 评论 复制链接分享
  • weixin_39942191 weixin_39942191 4月前

    The use-case for sub-directories is that they are automatically rebuild as soon as a dependency is updated. Let's assume a boost dependent pkg, if we increase the gloabl pinning on boost, all subdirs will be rebuild against the new boost version and hence stay compatible in larger environments.

    So subdirs are used for versions that the maintainer still wants to actively support and maintain.

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    I see. That makes sense.

    just fyi - someone just reported a bug in Trinity v2.8.2 that I'll need to resolve that wasn't encountered in any of the data sets I routinely test on, so we'll have a v2.8.3 in the not too distant future.

    ~b

    On Tue, Aug 21, 2018 at 11:32 AM Björn Grüning wrote:

    The use-case for sub-directories is that they are automatically rebuild as soon as a dependency is updated. Let's assume a boost dependent pkg, if we increase the gloabl pinning on boost, all subdirs will be rebuild against the new boost version and hence stay compatible in larger environments.

    So subdirs are used for versions that the maintainer still wants to actively support and maintain.

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bioconda/bioconda-recipes/pull/10425#issuecomment-414717349, or mute the thread https://github.com/notifications/unsubscribe-auth/AHMVX0Lqtuy_nmTg28l9m9qTKgc1P74uks5uTCgTgaJpZM4V8-DQ .

    --

    Brian J. Haas The Broad Institute http://broadinstitute.org/~bhaas http://broad.mit.edu/~bhaas

    点赞 评论 复制链接分享
  • weixin_39559369 weixin_39559369 4月前

    I just merged #10581, we'll see if the test of the galaxy tools pass with these changes. I guess we can wait 2.8.3 before merging https://github.com/galaxyproject/tools-iuc/pull/2037

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    sounds good. I'll keep you posted.

    On Tue, Aug 21, 2018 at 11:52 AM Anthony Bretaudeau < notifications.com> wrote:

    I just merged #10581 https://github.com/bioconda/bioconda-recipes/pull/10581, we'll see if the test of the galaxy tools pass with these changes. I guess we can wait 2.8.3 before merging galaxyproject/tools-iuc#203 https://github.com/galaxyproject/tools-iuc/pull/203

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bioconda/bioconda-recipes/pull/10425#issuecomment-414724351, or mute the thread https://github.com/notifications/unsubscribe-auth/AHMVX8N2_k2p_dAVLCeVA3Y2_7t4QfEHks5uTCzCgaJpZM4V8-DQ .

    --

    Brian J. Haas The Broad Institute http://broadinstitute.org/~bhaas http://broad.mit.edu/~bhaas

    点赞 评论 复制链接分享
  • weixin_39861823 weixin_39861823 4月前

    Thanks !

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    Trinity v2.8.3 is now out: https://github.com/trinityrnaseq/trinityrnaseq/releases/tag/Trinity-v2.8.3

    and so I'm hoping we can now complete the bioconda build. Please let me know if there's anything I can do to help.

    Thx in advance!

    ~brian

    点赞 评论 复制链接分享
  • weixin_39861823 weixin_39861823 4月前

    Thanks for the heads-up, see https://github.com/bioconda/bioconda-recipes/pull/10608

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    I can look into this.

    What's the best way for me to get to your github fork so I can put in a PR to you once it's ready?

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    yeah, this is pushing my knowledge of github.

    I've got a fork already... any guidance would be much appreciated. :-)

    点赞 评论 复制链接分享
  • weixin_39861823 weixin_39861823 4月前
    
    git remote add nsoranzo https://github.com/nsoranzo/bioconda-recipes
    git fetch nsoranzo
    git checkout trinity_2.8.3
    

    then make changes, commit and push as usual.

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    awesome! I'll give it a whirl.

    On Wed, Aug 22, 2018 at 11:54 AM Nicola Soranzo wrote:

    git remote add nsoranzo https://github.com/nsoranzo/bioconda-recipes git fetch nsoranzo git checkout trinity_2.8.3

    then make changes, commit and push as usual.

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bioconda/bioconda-recipes/pull/10425#issuecomment-415082643, or mute the thread https://github.com/notifications/unsubscribe-auth/AHMVXw_FvptyfdyxbbAZlnFJ5yC2NWnIks5uTX7JgaJpZM4V8-DQ .

    --

    Brian J. Haas The Broad Institute http://broadinstitute.org/~bhaas http://broad.mit.edu/~bhaas

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前
    • I was hoping to have everything work as we have it in the documentation, and have folks refer to specific scripts via the TRINITY_HOME environmental variable. In galaxy, it would be good to do this too. The bioconda installation makes a link to TRINITY_HOME in the opt/ area, and everything should be found relative to that path. Will that work for you?

    On Mon, Aug 20, 2018 at 10:44 AM Nicola Soranzo wrote:

    ** commented on this pull request.

    In recipes/trinity/build.sh https://github.com/bioconda/bioconda-recipes/pull/10425#discussion_r211288021 :

    -ln -s $TRINITY_HOME/util/misc/contig_ExN50_statistic.pl -cp $TRINITY_HOME/trinity-plugins/BIN/seqtk-trinity . - -# Find real path when executing from a symlink -export LC_ALL=C -find $TRINITY_HOME -type f -print0 | xargs -0 sed -i.bak 's/FindBin::Bin/FindBin::RealBin/g' - -# Replace hard coded path to deps -find $TRINITY_HOME -type f -print0 | xargs -0 sed -i.bak 's/$FASTOOL_DIR\/fastool/fastool/g' -find $TRINITY_HOME -type f -print0 | xargs -0 sed -i.bak 's/$JELLYFISH_DIR\/bin\/jellyfish/jellyfish/g' -find $TRINITY_HOME -type f -print0 | xargs -0 sed -i.bak 's/$COLLECTL_DIR\///g' -find $TRINITY_HOME -type f -print0 | xargs -0 sed -i.bak 's/${COLLECTL_DIR}\///g' -find $TRINITY_HOME -type f -print0 | xargs -0 sed -i.bak 's/$PARAFLY -c/ParaFly -c/g' -sed -i.bak "s/\$ROOTDIR\/trinity-plugins\/Trimmomatic/\/opt\/anaconda1anaconda2anaconda3\/share\/trimmomatic/g" $TRINITY_HOME/Trinity - -find $TRINITY_HOME -type f -name "*.bak" -print0 | xargs -0 rm -f

    https://github.com/brianjohnhaas It seems these changes removed all Trinity scripts from the PATH, which breaks at least the corresponding Galaxy wrappers (xref. galaxyproject/tools-iuc#2037 https://github.com/galaxyproject/tools-iuc/pull/2037 ). Can you revert them or is there a reason for the change?

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bioconda/bioconda-recipes/pull/10425#pullrequestreview-147676394, or mute the thread https://github.com/notifications/unsubscribe-auth/AHMVX2mT5TCbR3hGkPq9eeTKpQnZ8ZSyks5uSss6gaJpZM4V8-DQ .

    --

    Brian J. Haas The Broad Institute http://broadinstitute.org/~bhaas http://broad.mit.edu/~bhaas

    点赞 评论 复制链接分享
  • weixin_39942191 weixin_39942191 4月前

    I think we should make the tool accessible by default without knowing any detail. You are absolutely right in making the TRINITY_HOME variable use-case working, but I do not see any harm in putting them in PATH in addition.

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    We could put them back, but I was hoping to have everyone just use the TRINITY_HOME env var. There's a lot of other scripts that are sometimes used in Trinity as part of our documentation, and they'd still need to go through the TRINITY_HOME env var to find them from our documentation. Earlier, a subset of the scripts that are used from within galaxy were put in the bin area, I figured this was probably not improving maintainability. Any new galaxy app targeting a new Trinity application would need to then have the bioconda recipe include that new tool in the bin/ area, instead of just pointing to it through TRINITY_HOME. Second, it clutters the bin/ area by adding in a number of trinity scripts that may or may not get used by non-galaxy users. In short, it helps a subset of galaxy folks by doing this, but doesn't further benefit everyone else that's already using TRINITY_HOME.

    So - can we use TRINITY_HOME from now on? ;-) please...

    ~brian

    On Mon, Aug 20, 2018 at 11:56 AM Björn Grüning wrote:

    https://github.com/brianjohnhaas I think we should make the tool accessible by default without knowing any detail. You are absolutely right in making the TRINITY_HOME variable use-case working, but I do not see any harm in putting them in PATH in addition.

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bioconda/bioconda-recipes/pull/10425#issuecomment-414368395, or mute the thread https://github.com/notifications/unsubscribe-auth/AHMVX8-tlqe7WT5Rj9gAWuKzv_kY1YbNks5uStw3gaJpZM4V8-DQ .

    --

    Brian J. Haas The Broad Institute http://broadinstitute.org/~bhaas http://broad.mit.edu/~bhaas

    点赞 评论 复制链接分享
  • weixin_39958911 weixin_39958911 4月前

    I think we should follow the wishes of the package maintainer , but I have a few questions.

    A number of perl-* dependencies were removed. If those are optional dependencies, it's bioconda's general policy to include them to make the package easier to use if they actually are needed.

    There's no need to delete the recipes for the older versions. Especially for such a major change in the packaging, it's probably more helpful to leave them in for a period of time for projects depending on trinity to migrate to the new packaging methodology. For example, the 2.6.6 previous version can be archived in a 2.6.6 subfolder for reference. You can always delete the recipes at a later time.

    点赞 评论 复制链接分享
  • weixin_39683241 weixin_39683241 4月前

    Do we need to keep the recipes for the other packages if the build files are already available for them?

    I had deleted them for a couple reasons. First, they were on the blacklist and likely needed some additional maintenance updates so that they would build ok. Second, they were causing me some grief when I was trying to run 'conda build' and other testing locally. Then, there were lots of other older versions that were not included in the recipes but had conda build files for them, so I figured these would be fine to remove altogether as well.

    Disclaimer - I am relatively new to bioconda, and so still learning the ropes to a large extent. I'm happy to work with everyone to make the package more appealing to everyone else that's been contributing to it. Bioconda is becoming an important distribution mechanism for us, and I'd like to be an important part of it going forward though.

    best,

    ~brian

    On Mon, Aug 20, 2018 at 2:07 PM Simon Ye wrote:

    I think we should follow the wishes of the package maintainer https://github.com/brianjohnhaas, but I have a few questions.

    A number of perl-* dependencies were removed. If those are optional dependencies, it's bioconda's general policy to include them to make the package easier to use if they actually are needed.

    There's no need to delete the recipes for the older versions. Especially for such a major change in the packaging, it's probably more helpful to leave them in for a period of time for projects depending on trinity to migrate to the new packaging methodology. For example, the 2.6.6 previous version can be archived in a 2.6.6 subfolder for reference. You can always delete the recipes at a later time.

    — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/bioconda/bioconda-recipes/pull/10425#issuecomment-414410263, or mute the thread https://github.com/notifications/unsubscribe-auth/AHMVXyuYIEVkoMt89w_4X6jPLxvfRHWgks5uSvr7gaJpZM4V8-DQ .

    --

    Brian J. Haas The Broad Institute http://broadinstitute.org/~bhaas http://broad.mit.edu/~bhaas

    点赞 评论 复制链接分享

相关推荐