Dear list members, I have a question about memory limitations. I know the issue is frequently debated on the list, and I have checked the archive without finding the answers I need. I want to use AffylimmaGUI to analyse 80 cel files based on HGU133a chips from Affy. Initially, I tried to analyse the estrogen dataset and it worked very fine. Based on my own data I conducted a target file with the three column headlines Name FileName Target. However, when I open a new file and try to load the cel files throug the target file the loading process starts but then I recieve the message "Not enough memory". I have tried to change the memory allocated to R by the command memory.limit(size=2000). (I started wiht the size=200 and elevated the number.) i also tried to allocate more memory using the target file option under Properties. I use a 1.6 GHz Intel Pentium M processor, Windows XP and have 1 GB ram. My R version is R 1.9.1. Now, I consider what solution that will be the best for my future work wiht R, Bioconductor and limma:1) to upgrade my machine to 2 GB ram - would that be enough to handle my 80 cel files and what number of cel files will be the limit with 2 GB ram? 2) Alternatively, to buy a stationary computer: what would you recommend as a minimum size of processor and memory? In the long term I will have access to a continuous flow of cel files, so I want a solution that can handle more than "just" my present 80 cel files. 3) Are there any advantages to work on a unix system instead of a windows platform? I really hope some of you could give me some good advices before I start to convince my superviser that we need to make a big investment. Thanks in advance! Best regards, Malene Herbsleb Malene Herbsleb, MSc, Ph.D student Molecular Diagnostic Laboratory Aarhus University Hospital, Skejby Sygehus Brendstrupgaardsvej DK-8200 Aarhus N Denmark +45 89 49 51 29/ +45 89 42 31 31

Hi Malene, Sorry, affylmGUI is not very memory-efficient yet. Improving efficiency is certainly on my long list of things to do. I would suggest trying the justRMA function in the affy package from the R command-line or trying affylmGUI (and affy/justRMA) on a machine with more memory. For small affy analysis, you can use your own laptop or desktop, but for larger analyses e.g. 80 chips, you might want to try to gain access to a shared Linux server with more memory. Best regards, James

Malene, You will probably be better off going to linux, and depending on the number of chips you will be using in the future, you may even need to upgrade to a 64 bit chip in order to have sufficient RAM. Attached is a chart (slightly outdated) showing how much memory is required to read in various numbers of chips (32 bit windows and linux). An alternate view of the memory issue can be found here: http://stat-www.berkeley.edu/users/bolstad/ComputeRMAFAQ/size.html Best, Jim James W. MacDonald Affymetrix and cDNA Microarray Core University of Michigan Cancer Center 1500 E. Medical Center Drive 7410 CCGC Ann Arbor MI 48109 734-647-5623 >>> "Malene Herbsleb" <herbsleb@ki.au.dk> 08/09/04 06:52AM >>> Dear list members, I have a question about memory limitations. I know the issue is frequently debated on the list, and I have checked the archive without finding the answers I need. I want to use AffylimmaGUI to analyse 80 cel files based on HGU133a chips from Affy. Initially, I tried to analyse the estrogen dataset and it worked very fine. Based on my own data I conducted a target file with the three column headlines Name FileName Target. However, when I open a new file and try to load the cel files throug the target file the loading process starts but then I recieve the message "Not enough memory". I have tried to change the memory allocated to R by the command memory.limit(size=2000). (I started wiht the size=200 and elevated the number.) i also tried to allocate more memory using the target file option under Properties. I use a 1.6 GHz Intel Pentium M processor, Windows XP and have 1 GB ram. My R version is R 1.9.1. Now, I consider what solution that will be the best for my future work wiht R, Bioconductor and limma:1) to upgrade my machine to 2 GB ram - would that be enough to handle my 80 cel files and what number of cel files will be the limit with 2 GB ram? 2) Alternatively, to buy a stationary computer: what would you recommend as a minimum size of processor and memory? In the long term I will have access to a continuous flow of cel files, so I want a solution that can handle more than "just" my present 80 cel files. 3) Are there any advantages to work on a unix system instead of a windows platform? I really hope some of you could give me some good advices before I start to convince my superviser that we need to make a big investment. Thanks in advance! Best regards, Malene Herbsleb Malene Herbsleb, MSc, Ph.D student Molecular Diagnostic Laboratory Aarhus University Hospital, Skejby Sygehus Brendstrupgaardsvej DK-8200 Aarhus N Denmark +45 89 49 51 29/ +45 89 42 31 31 _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor

Doh! Here is the attachment..... James W. MacDonald Affymetrix and cDNA Microarray Core University of Michigan Cancer Center 1500 E. Medical Center Drive 7410 CCGC Ann Arbor MI 48109 734-647-5623 >>> "James MacDonald" <jmacdon@med.umich.edu> 08/09/04 08:51AM >>> Malene, You will probably be better off going to linux, and depending on the number of chips you will be using in the future, you may even need to upgrade to a 64 bit chip in order to have sufficient RAM. Attached is a chart (slightly outdated) showing how much memory is required to read in various numbers of chips (32 bit windows and linux). An alternate view of the memory issue can be found here: http://stat-www.berkeley.edu/users/bolstad/ComputeRMAFAQ/size.html Best, Jim James W. MacDonald Affymetrix and cDNA Microarray Core University of Michigan Cancer Center 1500 E. Medical Center Drive 7410 CCGC Ann Arbor MI 48109 734-647-5623 >>> "Malene Herbsleb" <herbsleb@ki.au.dk> 08/09/04 06:52AM >>> Dear list members, I have a question about memory limitations. I know the issue is frequently debated on the list, and I have checked the archive without finding the answers I need. I want to use AffylimmaGUI to analyse 80 cel files based on HGU133a chips from Affy. Initially, I tried to analyse the estrogen dataset and it worked very fine. Based on my own data I conducted a target file with the three column headlines Name FileName Target. However, when I open a new file and try to load the cel files throug the target file the loading process starts but then I recieve the message "Not enough memory". I have tried to change the memory allocated to R by the command memory.limit(size=2000). (I started wiht the size=200 and elevated the number.) i also tried to allocate more memory using the target file option under Properties. I use a 1.6 GHz Intel Pentium M processor, Windows XP and have 1 GB ram. My R version is R 1.9.1. Now, I consider what solution that will be the best for my future work wiht R, Bioconductor and limma:1) to upgrade my machine to 2 GB ram - would that be enough to handle my 80 cel files and what number of cel files will be the limit with 2 GB ram? 2) Alternatively, to buy a stationary computer: what would you recommend as a minimum size of processor and memory? In the long term I will have access to a continuous flow of cel files, so I want a solution that can handle more than "just" my present 80 cel files. 3) Are there any advantages to work on a unix system instead of a windows platform? I really hope some of you could give me some good advices before I start to convince my superviser that we need to make a big investment. Thanks in advance! Best regards, Malene Herbsleb Malene Herbsleb, MSc, Ph.D student Molecular Diagnostic Laboratory Aarhus University Hospital, Skejby Sygehus Brendstrupgaardsvej DK-8200 Aarhus N Denmark +45 89 49 51 29/ +45 89 42 31 31 _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor _______________________________________________ Bioconductor mailing list Bioconductor@stat.math.ethz.ch https://www.stat.math.ethz.ch/mailman/listinfo/bioconductor -------------- next part -------------- A non-text attachment was scrubbed... Name: Affy Memory Requirements.pdf Type: application/pdf Size: 67140 bytes Desc: not available Url : https://stat.ethz.ch/pipermail/bioconductor/attachments/20040809 /03cdcc9c/AffyMemoryRequirements.pdf