Create method to add more data (observation and samples)

[wasade]

perhaps Table.append and Table.extend with axis args?

[ElDeveloper]

Yeah that sounds like a good idea, follows the design of the rest of the API. BTW @adamrp @teravest and I were wondering why was this not a method before? Is there a reason for this?

On (Jun-15-14|20:48), Daniel McDonald wrote:

perhaps Table.append and Table.extend with axis args?

---

Reply to this email directly or view it on GitHub: https://github.com/biocore/biom-format/issues/475#issuecomment-46139042

[wasade]

Because there was merge On Jun 16, 2014 7:59 AM, "Yoshiki Vázquez Baeza" notifications@github.com wrote:

Yeah that sounds like a good idea, follows the design of the rest of the API. BTW @adamrp @teravest and I were wondering why was this not a method before? Is there a reason for this?

On (Jun-15-14|20:48), Daniel McDonald wrote:

perhaps Table.append and Table.extend with axis args?

---

Reply to this email directly or view it on GitHub: https://github.com/biocore/biom-format/issues/475#issuecomment-46139042

— Reply to this email directly or view it on GitHub https://github.com/biocore/biom-format/issues/475#issuecomment-46189331.

[wasade]

why not use a merge? I think a merge actually is the best thing to do here

[adamrp]

Yeah, that was my suggestion, too. On Jun 24, 2014 11:53 AM, "Daniel McDonald" notifications@github.com wrote:

why not use a merge?

— Reply to this email directly or view it on GitHub https://github.com/biocore/biom-format/issues/475#issuecomment-47005640.

[ElDeveloper]

It makes sense but from the UI perspective add would be more intuitive (and I can see add using merge under the hood without a problem).

Yoshiki Vázquez-Baeza

On Jun 24, 2014, at 11:53 AM, Daniel McDonald notifications@github.com wrote:

why not use a merge?

— Reply to this email directly or view it on GitHub.

[wasade]

What about:

orthogonal

def add(self, v, id_, md, orthogonal_ids, axis):
    if axis == 'sample':
        v = v[:, newaxis]
        to_merge = Table(v, orthogonal_ids, [id_], None, md)
    else:
        to_merge = Table(v, [id_], orthogonal_ids, md, None)
     return self.merge(to_merge)

[adamrp]

I still don't think this is worth the API change -- what is the use case? The use case that I heard about was for developers to create examples, but IMHO that's not a great reason to add it to the API, when we could almost as easily use a merge in that niche case.

On Thu, Jun 26, 2014 at 8:21 AM, Daniel McDonald notifications@github.com wrote:

https://camo.githubusercontent.com/708383c85740ce4d9e691d1fb7e918b14c93952f/687474703a2f2f322e62702e626c6f6773706f742e636f6d2f2d3577654b55335f394a62672f55764c55424147706530492f4141414141414141416a6f2f7073636b76755f67374d302f73313630302f636173746c652b7370656563686c6573732e676966

What about:

orthogonal

def add(self, v, id_, md, orthogonal_ids, axis): if axis == 'sample': v = v[:, newaxis] to_merge = Table(v, orthogonalids, [id], None, md) else: tomerge = Table(v, [id], orthogonal_ids, md, None) return self.merge(to_merge)

— Reply to this email directly or view it on GitHub https://github.com/biocore/biom-format/issues/475#issuecomment-47239639.

[wasade]

seems reasonable

On Thu, Jun 26, 2014 at 1:23 PM, adamrp notifications@github.com wrote:

I still don't think this is worth the API change -- what is the use case? The use case that I heard about was for developers to create examples, but IMHO that's not a great reason to add it to the API, when we could almost as easily use a merge in that niche case.

On Thu, Jun 26, 2014 at 8:21 AM, Daniel McDonald notifications@github.com

wrote:

< https://camo.githubusercontent.com/708383c85740ce4d9e691d1fb7e918b14c93952f/687474703a2f2f322e62702e626c6f6773706f742e636f6d2f2d3577654b55335f394a62672f55764c55424147706530492f4141414141414141416a6f2f7073636b76755f67374d302f73313630302f636173746c652b7370656563686c6573732e676966>

What about:

orthogonal

def add(self, v, id_, md, orthogonal_ids, axis): if axis == 'sample': v = v[:, newaxis] to_merge = Table(v, orthogonalids, [id], None, md) else: tomerge = Table(v, [id], orthogonal_ids, md, None) return self.merge(to_merge)

— Reply to this email directly or view it on GitHub https://github.com/biocore/biom-format/issues/475#issuecomment-47239639.

— Reply to this email directly or view it on GitHub https://github.com/biocore/biom-format/issues/475#issuecomment-47268286.

[wasade]

@ElDeveloper, okay to close?

[ElDeveloper]

No, I think we should have add as a method, however I don't think this should be there for 2.1.

[wasade]

Sounds good, setting to 2.2 and assigning you for now

[ElDeveloper]

Sounds good.

On (Jul-01-14|13:57), Daniel McDonald wrote:

Sounds good, setting to 2.2 and assigning you for now

---

Reply to this email directly or view it on GitHub: https://github.com/biocore/biom-format/issues/475#issuecomment-47708966

[ElDeveloper]

@wdwvt1 this is the issue that I mentioned during your code review (when you asked what was the easiest way to add data to a table) last week.

[wdwvt1]

Here was the use case (using merge now) I demonstrated in code review last week:

# remove features which don't meet criteria
fbt = bt.filter(_f, axis = 'observation', inplace = False)
# calculate the lost feature mass
lost_feature_mass = 1. - fbt.sum(axis = 'sample')
# create a new table with the lost feature mass as feature 'Other'
t = table.Table(lost_feature_mass, [lfm_name], fbt.ids(axis = 'sample'))
# merge tables and have lfm feature first to make plotting easier
return t.merge(fbt, sample = 'intersection', observation = 'union')

The merge accomplishes my objective just fine, but it seems silly to create a new table just for the purpose of adding a single feature.

[ElDeveloper]

The merge accomplishes my objective just fine, but it seems silly to create a new table just for the purpose of adding a single feature.

:+1:, I agree.

On (Jul-21-14|16:23), Will Van Treuren wrote:

Here was the use case (using merge now) I demonstrated in code review last week:

# remove features which don't meet criteria
fbt = bt.filter(_f, axis = 'observation', inplace = False)
# calculate the lost feature mass
lost_feature_mass = 1. - fbt.sum(axis = 'sample')
# create a new table with the lost feature mass as feature 'Other'
t = table.Table(lost_feature_mass, [lfm_name], fbt.ids(axis = 'sample'))
# merge tables and have lfm feature first to make plotting easier
return t.merge(fbt, sample = 'intersection', observation = 'union')

The merge accomplishes my objective just fine, but it seems silly to create a new table just for the purpose of adding a single feature.

---

Reply to this email directly or view it on GitHub: https://github.com/biocore/biom-format/issues/475#issuecomment-49679614

[wasade]

True. The Table provides the contextual information such as which element corresponds to what ID though. It would be a little ugly for an add method:

def add(self, id_, data, element_ids, axis='sample', metadata=None):
    ...

Where the element_ids need to be in index order with data, and be the corresponding IDs in the relevant axis of the Table. Do able, but still awkward. Would that be cleaner than the filter/merge stuff above?

[wasade]

@ElDeveloper, I think that #720 resolves this? Please reopen if necessary

[ElDeveloper]

👍!!

[ElDeveloper]

Though, just noticed the PR hasn't been merged 🤔

[wasade]

meh. Want to resolve that problem??? :)