Appendix 3:rma Algorithm Principle

The RMA algorithm is divided into three steps:

First, the background correction (no energy to write)

Second, normalization (no energy to write)

Third, calculate the expression value

Suppose there are 5 chips, some of which have 5 probes, and their expression values are as follows:

GeneChip

4 8 6) 9 7

3 1 2) 4 5

Probe 6 10 7 12 9

4 5 8) 9 6

7 11 8) 12 10

(1) Get the median of each line:

Rdelta

4 8 6 9 7 7

3 1 2 4 5 3

6 10 7 12 9 9

4 5 8 9 6 6

7 11 8 12 10 10

(2) Subtract this median from each line:

-3 1-1 2 0

0-2-1 1 2

-3 1-2 3 0

-2-1 2 3 0

-3 1-2 2 0

(3) Get the median of each column:

-3 1-1 2 0

0-2-1 1 2

-3 1-2 3 0

-2-1 2 3 0

-3 1-2 2 0

Cdelta-3 1-1 2 0

(4) Subtract this median from each column:

0 0 0) 0 0

3-3 0-1 2

0 0-1 1 0

1-2 3 1 0

0 0-1 0 0

(5) At this point, each row in (4) and The median of each column are 0. If this condition is not met, repeat (1)~(4) steps until they are met.

(6) Original matrix:

4 8 6) 9 7

3 1 2) 4 5

6 10 7) 12 9

4 5 8) 9 6

7 11 8) 12 10

Subtract (4) from the matrix:

0 0 0) 0 0

3-3 0-1 2

0 0-1 1 0

1-2 3 1 0

0 0-1 0 0

Get:

4 8 6) 9 7

0 4 2) 5 3

6 10 8) 11 9

3 7 5) 8 6

7 11 9) 12 10

Average for each column:

4 8 6) 9 7

--> the 5 number is the final result, the expression value of the probe group representing 5 samples

Appendix 3:rma Algorithm Principle