Hi Miao,

If you have the possibility to download the BAM files from cghub (you or your supervisors need to go through an approval process for it and ultimately you will receive a key file to used with gere torrent), you can just browse and select the normal samples here:

https://browser.cghub.ucsc.edu/

Normal sample are marked with 10 (Normal Blood) or 11 (Normal Tissue) in the barcode: TCGA-XX-YYYY-10Z-.... or  TCGA-XX-YYYY-11Z-....

Where XX-YYYY are the institute and patient ID, the rest of the barcode contains much more info (lookup at TCGA pages for it)

After selecting the samples you need you download a metadata.xml containing various file information and use it with GeneTorrent.

I don't think Level 2 data are available with normal sample information, but I might be wrong. So without permission to get Level 1 data there is no way to download patient data like germline variations and so on.

I hope this was more or less what you wanted to know.

I think "dbSNP" is what you are looking for. It will be probably be best to start with the wiki and click yourself onward from there: http://en.wikipedia.org/wiki/DbSNP

I have check the dbSNP at first time, the data provided in dbSNP are single mutation information and isolated from each other, so it's hard for me to make sure which mutation is sampled from normal people, and which mutations are from the same person(as the row of the input data, like the picture above).

So the kind of mutations I need, should not do damage to human body, but which doesn't mean those kind of mutations are only belong to Silent mutations, it allows those kind of mutations are other types, like Missense or Nonsense, just no harm to human or uncertain significant damage.

I am not sure this kind of mutations do exist certainly... So I come here ask for help.  If you have anything unclear, please feel free to contact me :)

Thanks.

Just to make sure I understand:

Are you looking for a collection of human sequencing data that contains all possible combinations of non-disease-causing SNPs? If this is the case, I am quite sure this kind of collection does not exist, and the number of problems involved trying to collect this data is so vast that I would not know where to start.

There are "control" genomes, for example from: http://www.1000genomes.org/data These will contain some, but by far not all possible non-disease-causing SNPs.

What is your aim? Could you please be more precise than "for a better understanding of the analysis of mutation genes in these certain cancer types"? Maybe you should read some reviews on these cancer types? Reviews often contains lists of all mutations known to cause a certain disease.

Hi Miao,

If you have the possibility to download the BAM files from cghub (you or your supervisors need to go through an approval process for it and ultimately you will receive a key file to used with gere torrent), you can just browse and select the normal samples here:

https://browser.cghub.ucsc.edu/

Normal sample are marked with 10 (Normal Blood) or 11 (Normal Tissue) in the barcode: TCGA-XX-YYYY-10Z-.... or  TCGA-XX-YYYY-11Z-....

Where XX-YYYY are the institute and patient ID, the rest of the barcode contains much more info (lookup at TCGA pages for it)

After selecting the samples you need you download a metadata.xml containing various file information and use it with GeneTorrent.

I don't think Level 2 data are available with normal sample information, but I might be wrong. So without permission to get Level 1 data there is no way to download patient data like germline variations and so on.

I hope this was more or less what you wanted to know.

Francesco's answer is better than mine and his link is a better source for your purposes, so please use his link!

Whether you go through the original BAM files or the list of mutations from the 1000 Genomes Project to obtain a set of non-pathologic variants, this will probably not solve your problem. Most cancer mutations are not driver mutations, but passenger mutations without a well-defined effect on the tissue, and there are currently no fool-proof ways to discern among bad mutations from neutral mutations in cancer datasets. Several groups, including ours, have used different criteria to filter out likely drivers, but these methods are empirical and subjective, at best. Nevertheless, you can follow one of the current methods to analyze your data (we used recurrent mutations, i.e. mutations at the same genomic coordinate in different patients, as a proxy for driver mutations, although this approach requires very large datasets).

Thanks all of you.

Francesco's answer gives me a hint, now I am trying to access data from that source and 1000genomes advised by Nicholas.

Also Thanks to Albino's advice,  Instead of filter divers mutations, I just want and try to use another method to find the gene of different expression among the cancer sample and normal sample. Worth a try.

hello miao,

I am trying to find the different expression gene in mRNA level among paired ESCC and normal sample.I have aquired the RNA-SEQ data from TCGA and the microarray data from NCBI. The problem is that the sample number is tooo small.Do you have any idea about the other access to the public data about paired mRNA expression data? 

Thank you.