SIAMCAT issueshttps://git.embl.de/grp-zeller/SIAMCAT/-/issues2020-01-31T14:43:09Zhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/107parameter for `evaluation.plot`2020-01-31T14:43:09ZJakob Wirbelparameter for `evaluation.plot`parameter to turn off ROC curves from individual cross-validation runsparameter to turn off ROC curves from individual cross-validation runsJakob WirbelJakob Wirbelhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/105special characters in label names are removed2019-10-14T12:16:10ZJakob Wirbelspecial characters in label names are removedspecial characters in label names are removed, so that the names will be the same if you use `(1, -1)` as labelsspecial characters in label names are removed, so that the names will be the same if you use `(1, -1)` as labelsJakob WirbelJakob Wirbelhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/103some minor issues with the multiple hypothesis testing in check.associations2019-10-14T12:16:01ZJakob Wirbelsome minor issues with the multiple hypothesis testing in check.associationsin `check.associations`, there are two adjustments to be made:
- allowed multiple hypothesis correction methods do not overlap with the allowed ones in `p.adjust`
- if no correction is performed, the plot should NOT say adj. p value in t...in `check.associations`, there are two adjustments to be made:
- allowed multiple hypothesis correction methods do not overlap with the allowed ones in `p.adjust`
- if no correction is performed, the plot should NOT say adj. p value in the x axisJakob WirbelJakob Wirbelhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/100bug in evaluate.predictions()2019-01-09T12:24:28ZKonrad Zychbug in evaluate.predictions()###Summary
evaluate predictions fails when `num.resample = 1`
###Relevant logs and/or screenshots
```
+ starting evaluate.predictions
+ calculating ROC
|==============================================================================...###Summary
evaluate predictions fails when `num.resample = 1`
###Relevant logs and/or screenshots
```
+ starting evaluate.predictions
+ calculating ROC
|====================================================================================| 100%
+ starting evaluate.classifier
+ finished evaluate.classifier
+ starting evaluate.get.pr
+ finished evaluate.get.pr
+ starting evaluate.calc.aupr
+ starting evaluate.get.pr
+ finished evaluate.get.pr
+ finished evaluate.calc.aupr
+ evaluating single prediction
Error in validObject(.Object) :
invalid class “eval_data” object: Entry for roc is not an object of class roc (from pROC)!
```
/cc @zych @jawirbelhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/99Bug in siamcat() creator function2019-01-09T12:22:50ZKonrad ZychBug in siamcat() creator function###Summary
siamcat() will always fail when otu_table has taxa in columns (it assumes rows)
/cc @zych @jawirbel###Summary
siamcat() will always fail when otu_table has taxa in columns (it assumes rows)
/cc @zych @jawirbelKonrad ZychKonrad Zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/97feature_weights/weight_matrix compute on the fly2018-09-07T11:33:41ZJakob Wirbelfeature_weights/weight_matrix compute on the fly### Summary
Instead of saving feature_weights/weight_matrix in the object, make the accessor functions compute them on the fly
### Affected functions/datasets/objects
`train.model`
`accessors`
/cc @zych### Summary
Instead of saving feature_weights/weight_matrix in the object, make the accessor functions compute them on the fly
### Affected functions/datasets/objects
`train.model`
`accessors`
/cc @zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/96on which type of features should the function work?2018-10-01T19:17:21ZJakob Wirbelon which type of features should the function work?### Summary
Include a parameter, on which type of features the function should work (i.e. original, filtered, normalized). Make smart defaults to keep the current workflow unaffected
### Affected functions/datasets/objects
- [x] `fil...### Summary
Include a parameter, on which type of features the function should work (i.e. original, filtered, normalized). Make smart defaults to keep the current workflow unaffected
### Affected functions/datasets/objects
- [x] `filter.features`
- [x] `normalize.features`
- [x] `check.associations`
- [x] `train.model`
- [x] `make.predictions`
- [x] `model.interpretation.plot`
- [x] `add.meta.pred`
- [x] `summarize.features`
/cc @zych @jawirbel https://git.embl.de/grp-zeller/SIAMCAT/-/issues/95remove meta-column after construction the label in the constructor function2018-09-21T08:57:59ZJakob Wirbelremove meta-column after construction the label in the constructor function### Summary
throw out the column with the label (for plotting reasons later)
only in the constructor function, not the `create.label` function
### Affected functions/datasets/objects
`siamcat-constructor`
/cc @zych### Summary
throw out the column with the label (for plotting reasons later)
only in the constructor function, not the `create.label` function
### Affected functions/datasets/objects
`siamcat-constructor`
/cc @zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/94Store feature weights in SIAMCAT2018-08-16T14:13:31ZJakob WirbelStore feature weights in SIAMCAT## Summary
Store mean/median feature weights in the slot for the models (for easier access, e.g. in `model.interpretation.plot`) and make them accessible by `model_weights(siamcat)` (or sth similar?)
## Affected functions/datasets/obj...## Summary
Store mean/median feature weights in the slot for the models (for easier access, e.g. in `model.interpretation.plot`) and make them accessible by `model_weights(siamcat)` (or sth similar?)
## Affected functions/datasets/objects
- `train.model`
- `model.interpretation.plot`
- `siamcat` Class
- `show` function
- accessors
- definition
/cc @zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/92easier label import2018-08-09T13:50:31ZJakob Wirbeleasier label importmake it possible to give a vector as label into the construction functionmake it possible to give a vector as label into the construction functionhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/91No NA handling in filter.features2018-08-09T13:52:14ZKonrad ZychNo NA handling in filter.features###Summary
If NAs are present in the data, the filter.features function does not work at all
###Relevant logs and/or screenshots
```
+ starting filter.features
+++ before filtering, the data has 1127 features
+++ applying abundance fi...###Summary
If NAs are present in the data, the filter.features function does not work at all
###Relevant logs and/or screenshots
```
+ starting filter.features
+++ before filtering, the data has 1127 features
+++ applying abundance filter
+++ checking for unmapped reads
+++ tried to remove unmapped reads, but could not find
them. Continue anyway.
+++ applying prune_taxa
+++ removed 1127 features whose values did not exceed 1e-04 in any sample (retaining 0)
Error in validObject(.Object) : invalid class "otu_table" object:
OTU abundance data must have non-zero dimensions.
```
/cc @zych @jawirbelKonrad ZychKonrad Zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/90clean the label class2018-08-01T14:14:26ZJakob Wirbelclean the label class- [x] remove redundant stuff in the label
- [x] fix all indexing calls with the label object- [x] remove redundant stuff in the label
- [x] fix all indexing calls with the label objecthttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/89fix constructor function2018-08-01T14:16:46ZJakob Wirbelfix constructor function- [x] use named arguments
- [x] perform some checks on features/labels/metadata
- [x] call data validator at the end of the constructor
- [x] SIAMCAT without the label (or NaN label)- [x] use named arguments
- [x] perform some checks on features/labels/metadata
- [x] call data validator at the end of the constructor
- [x] SIAMCAT without the label (or NaN label)Konrad ZychKonrad Zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/88bug in evaluate.predictions for models without resampling2018-07-19T13:36:17ZJakob Wirbelbug in evaluate.predictions for models without resampling### Summary
if no resampling has been performed (and thus the prediction is a single vector), the evaluation is buggy
### Relevant logs and/or screenshots
```
> + finished evaluate.predictions in 5.15 s
> Warning message:
> In pr[1]...### Summary
if no resampling has been performed (and thus the prediction is a single vector), the evaluation is buggy
### Relevant logs and/or screenshots
```
> + finished evaluate.predictions in 5.15 s
> Warning message:
> In pr[1] <- list(evaluate.get.pr(ev[[1]]), verbose = verbose) :
> number of items to replace is not a multiple of replacement length
```
/cc @zych @jawirbelhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/87Bug in model.interpretation.plot2018-07-19T12:07:04ZKonrad ZychBug in model.interpretation.plot###Summary
- [ ] the metadata display in the model interpretation plot looks a bit strange in the sense that some rows are very light throughout due to the smaller range of values - some form of normalization (rescale every row to the in...###Summary
- [ ] the metadata display in the model interpretation plot looks a bit strange in the sense that some rows are very light throughout due to the smaller range of values - some form of normalization (rescale every row to the interval [-1, 1])
/cc @zych @jawirbelhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/86Bugs in read.features2018-09-20T12:50:54ZKonrad ZychBugs in read.features### Summary
- [x] the read.features function doesn’t understand an URL as fn.in.feat (while e.g. read.table does)
- [ ] pass on arguments to the underlying read.table function
- [ ] make the function more flexible
/cc @zych @jawirbel### Summary
- [x] the read.features function doesn’t understand an URL as fn.in.feat (while e.g. read.table does)
- [ ] pass on arguments to the underlying read.table function
- [ ] make the function more flexible
/cc @zych @jawirbelKonrad ZychKonrad Zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/85Bugs in check.associations2018-07-24T15:38:17ZKonrad ZychBugs in check.associations### Summary
- [x] the prevalence shift panel in the association checking is buggy (sth is wrong with the y coordinate of the bars)
- [x] the legend in the quantile.rect plot is also misplaced - especially its title…
- [x] the sort.by op...### Summary
- [x] the prevalence shift panel in the association checking is buggy (sth is wrong with the y coordinate of the bars)
- [x] the legend in the quantile.rect plot is also misplaced - especially its title…
- [x] the sort.by option is either misleading or doesn’t work “fc” doesn’t produce a sorting by fold change...
/cc @zych @jawirbelhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/84Fix progressbars2019-10-14T09:28:42ZJakob WirbelFix progressbars### Summary
The messages following the progress bars are usually printed in the same line. Should be fixed. Also, progressbars should not appear in the vignette.
Example:
```{r}
+ finished filter.features in 0.038 s
|================...### Summary
The messages following the progress bars are usually printed in the same line. Should be fixed. Also, progressbars should not appear in the vignette.
Example:
```{r}
+ finished filter.features in 0.038 s
|===============================================================================================| 100%Trained lasso models successfully.
```
Proposed fix:
does not appear in `evaluate.predictions`, which also shows a progressbar from the `pROC` package. maybe we can try to do the progressbars in the same way...
### Affected Functions
- [ ] `make.predictions`
- [ ] `train.model`
- [ ] `check.associations`
/cc @zychKonrad ZychKonrad Zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/83Release 1.0.1 on bioconductor2018-07-23T13:58:14ZKonrad ZychRelease 1.0.1 on bioconductor### Summary
- [x] new vignettes
- [x] version bump
- [x] NEWS file
### NEWS file:
## Version 1.0.1
- updated function `create.label.from.metadata` to make it more flexible
- added `transpose` parameter to the `read.features` function
...### Summary
- [x] new vignettes
- [x] version bump
- [x] NEWS file
### NEWS file:
## Version 1.0.1
- updated function `create.label.from.metadata` to make it more flexible
- added `transpose` parameter to the `read.features` function
- added vignette describing holdout validation
- added vignette describing how to read in your data and process it with SIAMCAT
- added function `read.lefse` to read input files in LEfSe file format
- small bugfixes
/cc @zychBioconductor 1.0.1 releaseKonrad ZychKonrad Zychhttps://git.embl.de/grp-zeller/SIAMCAT/-/issues/82Better data validation2018-08-09T13:52:56ZJakob WirbelBetter data validation### Summary
- [ ] Check if the feature data consist of counts or of relative abundances. Throw an error if not
- [ ] Check if the label object is present and in the right format
- [ ] should go into the constructor function
- [ ] fix th...### Summary
- [ ] Check if the feature data consist of counts or of relative abundances. Throw an error if not
- [ ] Check if the label object is present and in the right format
- [ ] should go into the constructor function
- [ ] fix the case if features and labels have mutually exclusive samples
### Affected functions/datasets/objects
`read.features`
/cc @zychJakob WirbelJakob Wirbel