This plot will be more useful if your analyses are based on a
relatively small number of proteins. It represents the association, or
loading of each protein on the first two PCs. Longer arrows represents
-stronger associations. It is important to note that proteins that are
-weakly associated with PC1 or PC2 may still be highly correlated with
-other PCs not being plotted. Consult the scree plot (and other available
-methods) to determine the appropriate number of PCs to investigate.
+stronger associations.
+
+Note: Proteins that are weakly associated with PC1 or
+PC2 may still be highly correlated with other PCs not being plotted.
+Consult the scree plot (and other available methods) to determine the
+appropriate number of PCs to investigate.
+
visualize(dataPCA, graphType = "PCA_var", center = TRUE, scale = TRUE,
addlabels = TRUE)
-Any value may be selected here, if you prefer a different quantification schema, with the
-exception of Total Spectra or Weighted Spectra. Do not use these.
+**Note:** Any value may be selected here, if you prefer a different quantification schema,
+with the exception of Total Spectra or Weighted Spectra. Do not use these.
### 2. Turning Off Normalization in Scaffold
@@ -56,11 +56,11 @@ normalize by various methods in msDiaLogue, but you should not stack normalizati
between programs.
-Non-quantifiable abundance values are reported in **Scaffold** as 0 or 1, and msDiaLogue
-`preprocessing` knows to treat these as NA. If normalization is applied in **Scaffold**,
-non-quantifiable abundances are transformed into fractional values which will not be
-converted properly in `preprocessing` and your data will no longer yield sensible
-analyses.
+**Note:** Non-quantifiable abundance values are reported in **Scaffold** as 0 or 1, and
+msDiaLogue `preprocessing` knows to treat these as NA. If normalization is applied in
+**Scaffold**, non-quantifiable abundances are transformed into fractional values which
+will not be converted properly in `preprocessing` and your data will no longer yield
+sensible analyses.
### 3. Minimum Requirements for Experiment Conditions
@@ -71,9 +71,9 @@ replicates for any condition, or having only 1 condition, will throw an error in
msDiaLogue and you will not be able to process your data.
-You can't sensibly estimate sample variance from 2 replicates, so statistical tests really
-should not be run on fewer than 3 replicates in general, whether in msDiaLogue or other
-programs.
+**Note:** You can't sensibly estimate sample variance from 2 replicates, so statistical
+tests really should not be run on fewer than 3 replicates in general, whether in
+msDiaLogue or other programs.
### 4. Sample Naming Format Requirements
@@ -86,9 +86,9 @@ the tab of each sample individually, right-clicking the tab, choosing "Edit BioS
typing the correct name format into the "Sample Name" box, and clicking "Apply".
-**Scaffold 5** will send you back to the "Samples" tab each time "Apply" is clicked, so
-this step is a little tedious to do in **Scaffold**. You can also fix this in **Excel**
-after exporting, before you import into **R**, which will likely be easier.
+**Note:** **Scaffold 5** will send you back to the "Samples" tab each time "Apply" is
+clicked, so this step is a little tedious to do in **Scaffold**. You can also fix this in
+**Excel** after exporting, before you import into **R**, which will likely be easier.
### 5. Filtering Dataset for Protein Identification
@@ -97,7 +97,7 @@ peptide identified. In the "Samples" tab, at the top-most menu bar, in the "Min
dropdown, set this to 2.
-This is not required, and if you choose to evaluate the dataset with 1-peptide
+**Note:** This is not required, and if you choose to evaluate the dataset with 1-peptide
identifications included, msDiaLogue will not stop you. However, the **Scaffold** report
does not provide information about how many peptides/protein were identified, so unlike
with a **Spectronaut**-based report, you cannot filter based on this information once in
@@ -136,10 +136,11 @@ specify the `fileName` to read the raw data file into **R**.
first load the data file directly, then specify the `dataSet` in the function.
-Decoys and contaminants can be removed either in the **Excel** report before Preprocessing,
-or can be removed in msDiaLogue with the filter step, if the accession number has the
-"CON__" or "DECOY__" prefix. (This prefix is not applied with search algorithms like
-MSFragger, so alternative filters would need to be developed in that case.)
+**Note:** Decoys and contaminants can be removed either in the **Excel** report before
+Preprocessing, or can be removed in msDiaLogue with the filter step, if the accession
+number has the "CON__" or "DECOY__" prefix. (This prefix is not applied with search
+algorithms like MSFragger, so alternative filters would need to be developed in that
+case.)
+**Note:** Not all proteins can be identified unambiguously. In many cases, the identified
peptides can be found in multiple protein sequences, which yields a protein group or
protein cluster rather than a single protein identification. When this happens, the
accession numbers for all potential matches are concatenated into one string, separated by
periods. When you see long strings of multiple identifiers later in your data processing,
this is why. Spectronaut sorts these alphanumerically, so you should not assume that the
first protein in the list is most likely to be correct (as is the case in other search
-algorithms).
+algorithms).
**5\.** Saves a document to your working directory with all filtered out data, if desired
@@ -118,8 +119,9 @@ dataSet <- preprocessing(fileName,
replaceBlank = TRUE, saveRm = TRUE)
```
-**NOTE:** `preprocessing()` does not perform a transformation on your data. You still need
-to use the function `transform()`.
+
+**Note:** `preprocessing()` does not perform a transformation on your data. You still need
+to use the function `transform()`.
```{r}
## if the raw data is in an .Rdata file
@@ -167,12 +169,12 @@ researcher wants to evaluate data from only one species at a time. This step all
set aside specific proteins from downstream analysis, using the gene_species identifier
format.
-**NOTE:** The proteins to be selected or removed is the union of those specified in
-`listName` and those matching the regular expression pattern in `regexName`.
-
-**KEEP IN MIND:** Removal of any proteins, including common contaminants, will affect any
+
+**Note:** The proteins to be selected or removed is the union of those specified in
+`listName` and those matching the regular expression pattern in `regexName`. \
+**Keep in mind:** Removal of any proteins, including common contaminants, will affect any
global calculations performed after this step (such as normalization). This should not be
-done without a clear understanding of how this will affect your results.
+done without a clear understanding of how this will affect your results.
**Case 1.** Remove proteins specified by the user in this step and keep everything else.
@@ -397,13 +399,14 @@ For example, to impute the NA value of `dataNorm` using `impute.min_local()`, se
required percentage of values that must be present in a given protein by condition
combination for values to be imputed to 51\%.
-**NOTE:** There is no rule in the field of proteomics for filtering based on percentage of
+
+**Note:** There is no rule in the field of proteomics for filtering based on percentage of
missingness, similar to there being no rule for the number of replicates required to draw
a conclusion. However, reproducible observations make conclusions more credible. Setting
the `reqPercentPresent` to 0.51 requires that any protein be observed in a majority of the
replicates by condition in order to be considered. For 3 replicates, this would require 2
-measurements to allow imputation of the 3rd value. If only 1 measurement is seen, the other
-values will remain NA, and will be filtered out in a subsequent step.
+measurements to allow imputation of the 3rd value. If only 1 measurement is seen, the
+other values will remain NA, and will be filtered out in a subsequent step.
```{r}
dataImput <- impute.min_local(dataNorm, reportImputing = FALSE,
@@ -643,9 +646,10 @@ variance." This clearly occurs when columns representing proteins contain only z
have constant values. Typically, there are two ways to address this error: one is to
remove these proteins, and the other is to set `scale = FALSE`.
-**NOTE:** Data scaling is done to ensure that the scale differences between different
+
+**Note:** Data scaling is done to ensure that the scale differences between different
features do not affect the results of PCA. If not scaled, features with larger scales will
-dominate the computation of principal components (PCs).
+dominate the computation of principal components (PCs).
In the case of `dataImput`, two proteins, namely "TEBP_HUMAN" and "T126B_HUMAN,"
have constant values, leading to the error message. We choose to remove these two proteins
@@ -703,10 +707,12 @@ visualize(dataPCA, graphType = "PCA_ind", center = TRUE, scale = TRUE,
This plot will be more useful if your analyses are based on a relatively small number of
proteins. It represents the association, or loading of each protein on the first two PCs.
-Longer arrows represents stronger associations. It is important to note that proteins that
-are weakly associated with PC1 or PC2 may still be highly correlated with other PCs not
-being plotted. Consult the scree plot (and other available methods) to determine the
-appropriate number of PCs to investigate.
+Longer arrows represents stronger associations.
+
+
+**Note:** Proteins that are weakly associated with PC1 or PC2 may still be highly
+correlated with other PCs not being plotted. Consult the scree plot (and other available
+methods) to determine the appropriate number of PCs to investigate.
```{r}
visualize(dataPCA, graphType = "PCA_var", center = TRUE, scale = TRUE,