I am looking for a publicly available matched tumor-normal sample. I need Illumina fastq reads (or an aligned bam file, since I could extract the reads from it) from a tumor and a matching, non-tumor control set from the same patient. Ideally, I would also have a truth set of known variants in the sample, but I can live without that.

I checked the Next Generation Cancer Knowledge Network but although they do seem to have appropriate bam files, which I could convert to fastq, all of the ones I found were "controlled access" and getting access seems very complicated.

Is there something like the public NA12878 sample for tumor-normal analyses?

  • 1
    $\begingroup$ The TCGA does provide the raw fastq too, and for some patients one has sample from the tumor and from a healthy region. But I am not sure it is public as without registration $\endgroup$ – llrs Apr 20 '18 at 15:11
  • $\begingroup$ @Llopis as far as I can tell, the registration process is incredibly convoluted and involves getting the "Signing officer" of the organization (PI, CEO, whatever) to register the organization with two separate governmental bodies (eRA and then dbGAP) and then ask for access. It might be possible for me to ask my boss to do this, but I would much prefer to find more easily accessible data. $\endgroup$ – terdon Apr 20 '18 at 15:37
  • 1
    $\begingroup$ Most patient data is stored in dbGAP because of privacy issues, so you'll struggle to find anything that's unrestricted. $\endgroup$ – heathobrien Apr 20 '18 at 15:49
  • $\begingroup$ @heathobrien yeah, which is why I've been struggling :) I was hoping there was a Henrietta Lacks analogue for the modern day or that NIST had released a tumor-normal pair or something. $\endgroup$ – terdon Apr 20 '18 at 15:54
  • $\begingroup$ I thought the HeLa genome got pulled down for exactly that reason. Is it available again? $\endgroup$ – heathobrien Apr 20 '18 at 16:03

Please take a look at

An open access pilot freely sharing cancer genomic data from participants in Texas

Although I have not worked on this paper but author claimed that both tumor and normal data is publicly available. In case if you still need something more, than you can work on virtual normal data. And there are several papers which provide machnine learning techniques to overcome the deficiency of normal data (not much efficient but still they give considerable results). Hope this information will help you. Thanks

  • $\begingroup$ Argh, this looks perfect, but the site seems to be down! I have emailed the author of the paper to see if it's just a temporary glitch. Thanks! $\endgroup$ – terdon Apr 25 '18 at 13:43
  • $\begingroup$ The site is no longer maintained, but I contacted the authors and they very helpfully put it back up for me! $\endgroup$ – terdon Apr 26 '18 at 23:35
  • $\begingroup$ @terdon: Then you should look for virtual normal database than or, if interested, machine learning techniques. $\endgroup$ – Lot_to_learn Apr 30 '18 at 6:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.