...

...

@@ 617,16 +617,16 @@ negative (lower is better) signature.

617

617

expression PCs.

618

618

Positive signature.

619

619

* EXP_QC_COR: Removal of Alignment Artifacts.

620


 Maximum squared Spearman correlation between first 3 expression PCs and


620

+ R^2 measure for regression of first 3 expression PCs on

621

621

first `k_qc` QPCs.

622

622

Negative signature.

623

623

* EXP_UV_COR: Removal of Expression Artifacts.

624


 Maximum squared Spearman correlation between first 3 expression PCs and


624

+ R^2 measure for regression of first 3 expression PCs on

625

625

first 3 PCs of the negative control (specified by `eval_negcon` or

626

626

`ruv_negcon` by default) submatrix of the original (raw) data.

627

627

Negative signature.

628

628

* EXP_WV_COR: Preservation of Biological Variance.

629


 Maximum squared Spearman correlation between first 3 expression PCs and


629

+ R^2 measure for regression of first 3 expression PCs on

630

630

first 3 PCs of the positive control (specified by `eval_poscon`)

631

631

submatrix of the original (raw) data.

632

632

Positive signature.

...

...

@@ 648,7 +648,7 @@ my_scone < scone(my_scone,

648

648

run=TRUE,

649

649

eval_kclust = 2:6,stratified_pam = TRUE,

650

650

return_norm = "in_memory",

651


 zero = "preadjust")


651

+ zero = "postadjust")

652

652


653

653

```

654

654


...

...

@@ 665,9 +665,9 @@ In the call above, we have set the following parameter arguments:

665

665

Store all normalized matrices in addition to evaluation data.

666

666

Otherwise normalized data is not returned in the resulting

667

667

object.

668


* zero = "preadjust".

669


 Restore data entries that are originally zeroes back to zero

670


 after the scaling step.


668

+* zero = "postadjust".


669

+ Restore data entries that are originally zeroes / negative after


670

+ normalization to zero after the adjustment step.

671

671


672

672

The output will contain various updated elements:

673

673


...

...

@@ 695,7 +695,7 @@ be recomputed.

695

695

# Step 3: Selecting a normalization for downstream analysis

696

696


697

697

Based on our sorting criteria, it would appear that

698


`none,fq,no_uv,no_bio,no_batch` performs well compared to other normalization


698

+`none,uq,ruv_k=1,no_bio,no_batch` performs well compared to other normalization

699

699

workflows. A useful way to visualize this method with respect to others is the

700

700

`biplot_color` function

701

701

