[BioC] LIMMA: Unbalanced technical replicates

[Gordon K Smyth]

> Date: Mon, 09 May 2005 16:34:02 +0300
> From: "Ron Ophir" <ron.ophir at weizmann.ac.il>
> Subject: [BioC] LIMMA: Unbalanced technical replicates
> To: <bioconductor at stat.math.ethz.ch>
>
> Hi,
> I have an affymetrix experiment with the following target file:
> Replicates	FileName	Target
> 8	SA100Hu133plus.CEL	LR
> 4	SA101Hu133plus.CEL	LR
> 4	SA102Hu133plus.CEL	LR
> 4	SA103Hu133plus.CEL	LR
> 9	SA104Hu133plus.CEL	LR
> 5	SA105Hu133plus.CEL	LR
> 5	SA106Hu133plus.CEL	LR
> 10	SA107Hu133plus.CEL	LR
> 11	SA108Hu133plus.CEL	LR
> 12	SA109Hu133plus.CEL	MET
> 13	SA111Hu133plus.CEL	MET
> 14	SA112Hu133plus.CEL	MET
> 15	SA113Hu133plus.CEL	MET
> 16	SA114Hu133plus.CEL	MET
> 17	SA115Hu133plus.CEL	MET
> 18	SA116Hu133plus.CEL	MET
> 6	SA117Hu133plus.CEL	MET
> 6	SA119Hu133plus.CEL	MET
> 19	SA121Hu133plus.CEL	METD
> 20	SA122Hu133plus.CEL	METD
> 7	SA123Hu133plus.CEL	METD
> 7	SA124Hu133plus.CEL	METD
> 21	SA127Hu133plus.CEL	HR
> 1	SA83Hu133plus.CEL	HR
> 1	SA84Hu133plus.CEL	HR
> 1	SA85Hu133plus.CEL	HR
> 22	SA86Hu133plus.CEL	HR
> 2	SA87Hu133plus.CEL	HR
> 2	SA88Hu133plus.CEL	HR
> 2	SA89Hu133plus.CEL	HR
> 23	SA90Hu133plus.CEL	HR
> 3	SA92Hu133plus.CEL	HR
> 3	SA93Hu133plus.CEL	HR
> 24	SA94Hu133plus.CEL	HR
> 25	SA95Hu133plus.CEL	HR
> 26	SA96Hu133plus.CEL	HR
> 27	SA98Hu133plus.CEL	LR
> 28	SA99Hu133plus.CEL	LR
> where files with the same number under replicates column are blocks of
> the technical replicates of the sames biological replicate.
> and the following design:
> 	LR	HR	MET	METD
> SA100Hu133plus.CEL	1	0	0	0
> SA101Hu133plus.CEL	1	0	0	0
> SA102Hu133plus.CEL	1	0	0	0
> SA103Hu133plus.CEL	1	0	0	0
> SA104Hu133plus.CEL	1	0	0	0
> SA105Hu133plus.CEL	1	0	0	0
> SA106Hu133plus.CEL	1	0	0	0
> SA107Hu133plus.CEL	1	0	0	0
> SA108Hu133plus.CEL	1	0	0	0
> SA109Hu133plus.CEL	0	0	1	0
> SA111Hu133plus.CEL	0	0	1	0
> SA112Hu133plus.CEL	0	0	1	0
> SA113Hu133plus.CEL	0	0	1	0
> SA114Hu133plus.CEL	0	0	1	0
> SA115Hu133plus.CEL	0	0	1	0
> SA116Hu133plus.CEL	0	0	1	0
> SA117Hu133plus.CEL	0	0	1	0
> SA119Hu133plus.CEL	0	0	1	0
> SA121Hu133plus.CEL	0	0	0	1
> SA122Hu133plus.CEL	0	0	0	1
> SA123Hu133plus.CEL	0	0	0	1
> SA124Hu133plus.CEL	0	0	0	1
> SA127Hu133plus.CEL	0	1	0	0
> SA83Hu133plus.CEL	0	1	0	0
> SA84Hu133plus.CEL	0	1	0	0
> SA85Hu133plus.CEL	0	1	0	0
> SA86Hu133plus.CEL	0	1	0	0
> SA87Hu133plus.CEL	0	1	0	0
> SA88Hu133plus.CEL	0	1	0	0
> SA89Hu133plus.CEL	0	1	0	0
> SA90Hu133plus.CEL	0	1	0	0
> SA92Hu133plus.CEL	0	1	0	0
> SA93Hu133plus.CEL	0	1	0	0
> SA94Hu133plus.CEL	0	1	0	0
> SA95Hu133plus.CEL	0	1	0	0
> SA96Hu133plus.CEL	0	1	0	0
> SA98Hu133plus.CEL	1	0	0	0
> SA99Hu133plus.CEL	1	0	0	0
> My question is should I run duplicateCorrelation function with this
> vector
> B<-c( 8,  4,  4,  4,  9,  5,  5, 10, 11, 12, 13, 14, 15, 16, 17, 18,
> 6,  6, 19, 20,  7,  7, 21,  1,  11, 22,  2,  2,  2, 23,  3,  3, 24, 25,
> 26, 27, 28)
> where each biological replicate gets different block number

Yes.

> or
> B<-c(   4,  4,  4,   5,  5,  6,  6,   7,  7,   2,  2,  2, 3,  3, )
> where I ignore biological replicates or maybe put zero instead of
> biological replicates.

No.

> corfit <- duplicateCorrelation(data, design, block=B)

'block=targets$Replicate' would be fine.

Gordon

> Thanks
> Ron