Roadmap

[jibarozzo]

Focusing on
extract_core()
scnm.fit()
multi_rarefy()
parallel_rarefy()
parallel_extract_core()
visualization function

Vision:

parallel_extract_core() now identify_core()

• focus on parallelization (@jibarozzo)

*_rarefy()

• Remove rarefaction, give rarefaction option to end-user

• sequential and parallel implementation (@Gian77 @jibarozzo)

• summarized and non-summarized results (@jibarozzo )

scnm.fit()

• get it to work (@Gian77)

viz_function

• make function @Gian77

original dataset

• add the original dataset form Shade and Stopnisek @Gian77 as in original dataset for benchmarking #52

[Gian77]

New PR #53 will add

• Three datasets (switchgrass.rda, mimulus.rda, bean.rda) used by Shade and Stopnisek (2019) Curr Opin Biol [1]
• identify_core() function with the possibility to weight the OTU occurrence by abundances ( as mentioned in [1] but I think never implemented, now we have it and can explian how it works)
• plot_identified_core() which plot the % Bracy-Curtis increase over the ranked OTUs and marks the elbow and last % increase cutoffs points.

Net steps are:

• cleanup not used plotting functions ( all those related to PCoA and NMDS... @jibarozzo)
• not-needed auxiliary functions (e.g. calculate_bc() @jibarozzo )
• possibly old datasets? But we do need a dataset with many samples for testing and further development.

In general, @jibarozzo we should decide what to keep and what not. I guess if we are following Shade and Stopnisek we should stick to what they provided plus a few helper functions like the multi_rarefy() etc.

[Gian77]

With the new PR #55 I am removing multi_rarefy.R and model_fit.R since are redundant and we have them in the experimental branch.

[Gian77]

All of the above is changed I think we should close the first message of the issue since we modified everything basically...

Something to work now on:

test_phyloseq

• is works but it seems weird to me, some OTU/ASVs have very low occurrence and high abundance and they are very similar one another (contaminants?)

Just noticed we have a problem with plot_neutral_model() the taxa not neutral and not core are not displayed, will fix asap.

• we should rename occupancy_abundance into abundance_occupancy in any place they appear for consistency.

Carried over from previous messages

parallel_extract_core() now identify_core()

• focus on parallelization (@jibarozzo)

*_rarefy() I think this is good now...

• Remove rarefaction, give rarefaction option to end-user

• sequential and parallel implementation (@Gian77 @jibarozzo)

• summarized and non-summarized results (@jibarozzo )

Also keep in mind #56 to polish this as best before a future launch

[jibarozzo]

@Gian77 Looks really good. Yeah, maybe those weird OTUs are contaminants or controls. I don't know the dataset well. I think test-phyloseq.rda is focused on swithgrass.