Add calc_prop_samples

[zachary-foster]

This would be like calc_n_samples except it would return the proportion of samples in each group with greater than some number of reads.

[zachary-foster]

done

[olar785]

Hi Zachary,
I have been trying the calc_prop_samples function (which is very useful by the way) and there seems to be a problem with the labelling of groups returned. In my case I had 3 groups of samples (Nodule, Sediment, Water) and I realised that the values reported for Sediment and Water were inverted. Using the same input data I calculated the values myself for some specific taxa and could confirm that the values for Sediment and Water samples were inverted. I can send you my data if you need.
Thanks

[zachary-foster]

Hi @olar785, thanks for the report. Did you use the cols options to specify the columns to use? Something like that might happen if you don't use the cols option and the numeric columns in the table are in a different order than the sample type given to the groups option. Perhaps I should add a warning about that. If you did use the cols option (and it was in the same order as the values given to the group option), then it sounds like a bug. If that is the case, can you send me you data and code? Thanks!

[olar785]

Thanks for your quick reply. I did use the cols option. Here is what I did: taxmap_18S$data$propS = calc_prop_samples(taxmap_18S, "tax_abund", cols = taxmap_18S$data$sample_data$sample_id, groups = taxmap_18S$data$sample_data$Sample_type, other_cols= TRUE) However, if I use the compare_groups() functions with the same parameters I get the results I expect. In this case for example, I can see that the Cnidaria phylum is much more present in the water samples than in the sediment ones: taxmap_18S$data$diff_table <- compare_groups(taxmap_18S, "tax_abund", cols = taxmap_18S$data$sample_data$sample_id, groups = taxmap_18S$data$sample_data$Sample_type) I have attached my data below. Thanks

…

On 30/01/2019, at 8:03 AM, Zachary Foster ***@***.***> wrote: Hi @olar785 <https://github.com/olar785>, thanks for the report. Did you use the cols options to specify the columns to use? Something like that might happen if you don't use the cols option and the numeric columns in the table are in a different order than the sample type given to the groups option. Perhaps I should add a warning about that. If you did use the cols option (and it was in the same order as the values given to the group option), then it sounds like a bug. If that is the case, can you send me you data and code? Thanks! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#233 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ATdp7XN7rpGUR9tsVIKyLRpipAcezwDnks5vId5agaJpZM4T5F99>.

[zachary-foster]

Ok, I will look into it. I don't see your attachment though. Perhaps you need to attach it on Github rather than via email? Thanks

[zachary-foster]

You can also email it to me directly at "[email protected]" if you would rather it not be posted on github

[zachary-foster]

Hi @olar785, it was a bug and I think it is fixed now. I did not notice it before because it only effects some datasets when the groups option is used and the bug did not affect my test data set. Try installing the development version and let me know if it works for you:

devtools::install_github("grunwaldlab/metacoder")

> taxmap_18S <- readRDS("~/Downloads/taxmap_18S.rds")
> 
> hand_res <- lapply(unique(taxmap_18S$data$sample_data$Sample_type), function(site) {
+   site_ids <- taxmap_18S$data$sample_data$sample_id[taxmap_18S$data$sample_data$Sample_type == site]
+   site_data <- taxmap_18S$data$tax_abund[site_ids]
+   hand_res <- apply(MARGIN = 1, site_data, function(x) sum(x > 0) / length(x))
+ })
> names(hand_res) <- unique(taxmap_18S$data$sample_data$Sample_type)
> hand_res <- tibble::as_tibble(hand_res)
> 
> res <- calc_prop_samples(taxmap_18S, "tax_abund", cols = taxmap_18S$data$sample_data$sample_id, groups = taxmap_18S$data$sample_data$Sample_type, other_cols= TRUE)
Calculating the proportion of samples with a value greater than 0 for 219 columns in 3 groups for 556 observations
Warning message:
The following columns will be replaced in the output:
   N-1, N-10, N-11, N-12, N-13, N-14, N-15, N-16, N-17, N-18, N-19, N-2 ... S-216, S-219, S-230, S-241, S-306, S-46, S-48, S-54, S-56, S-87, S-98
 
> 
> all(res[,-1] == hand_res)
[1] TRUE

Thanks again for the report!

[olar785]

Hi Zachary, I removed the metacoder package and tried to reinstall it with devtools::install_github(“grunwaldlab/metacoder”) but I get the following error: Error: package or namespace load failed for ‘metacoder’ in namespaceExport(ns, exports): undefined exports: edge_shape_functions, node_shape_functions I don’t if it’s an issue stemming from my configuration or something else though.

…

On 30/01/2019, at 1:17 PM, Zachary Foster ***@***.***> wrote: Hi @olar785 <https://github.com/olar785>, it was a bug and I think it is fixed now. I did not notice it before because it only effects some datasets when the groups option is used and the bug did not affect my test data set. Try installing the development version and let me know if it works for you: devtools::install_github("grunwaldlab/metacoder") > taxmap_18S <- readRDS("~/Downloads/taxmap_18S.rds") > > hand_res <- lapply(unique(taxmap_18S$data$sample_data$Sample_type), function(site) { + site_ids <- taxmap_18S$data$sample_data$sample_id[taxmap_18S$data$sample_data$Sample_type == site] + site_data <- taxmap_18S$data$tax_abund[site_ids] + hand_res <- apply(MARGIN = 1, site_data, function(x) sum(x > 0) / length(x)) + }) > names(hand_res) <- unique(taxmap_18S$data$sample_data$Sample_type) > hand_res <- tibble::as_tibble(hand_res) > > res <- calc_prop_samples(taxmap_18S, "tax_abund", cols = taxmap_18S$data$sample_data$sample_id, groups = taxmap_18S$data$sample_data$Sample_type, other_cols= TRUE) Calculating the proportion of samples with a value greater than 0 for 219 columns in 3 groups for 556 observations Warning message: The following columns will be replaced in the output: N-1, N-10, N-11, N-12, N-13, N-14, N-15, N-16, N-17, N-18, N-19, N-2 ... S-216, S-219, S-230, S-241, S-306, S-46, S-48, S-54, S-56, S-87, S-98 > > all(res[,-1] == hand_res) [1] TRUE — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#233 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ATdp7atrnUZVmMOP70OPKDC3L6_O69Usks5vIigSgaJpZM4T5F99>.

[zachary-foster]

Sorry about that @olar785, I was working on the development branch when I made that bug fix to the master branch and did not rebuild the namespace after switching branches to commit the bug fix on master, so there were references in the namespace to functions that did not exist in master. It should be working now. At least it works for me now.

[olar785]

Works like a charm, thank you!

[zachary-foster]

no problem!