Blurry Lines and the Cost Curve
27/03/12 01:54 Filed in: GenomeWeb Daily Scan
Submitted by S. Pelech - Kinexus on Tue, 03/27/2012 - 01:54.
At Kinexus Bioinformatics Corporation, we have recently completed a detailed meta-analysis of all of the reported mutations in over 3000 human genes that have been linked with cancer in one way or another. Much of the data was retrieved from the Sanger Institute's Catalogue of Somatic Mutations in Cancer (COSMIC) database, which contains over 217,000 mutations described for over 683,000 tumours. While there are some instances of well defined cancer-related proteins such as Ras isoforms and isocitrate dehydrogenase with clear hotspots of mutation, the vast majority feature mutations that are apparently randomly distributed throughout their sequences, and in most cases less than 1% of their amino acids have been observed to be altered with any cancer. In fact, for more than half of these cancer-associated proteins, either none or only 1 or 2 mutations have been discovered. In retrospect, this is not surprising, since gene mutation by radiation and chemical carcinogens is a random process, and a hallmark of advanced cancers is the accumulation of sporadic mutations due to damage to proteins that repair DNA.
With some 60 million single nucleotide polymorphisms likely to exist within the human genome, it will be some time before we can sort through all of the data and identify most of the bona fide pathologically relevant mutations in the human genome. This is certainly exemplified with the difficulty associated with assignment of driver and bystander mutations in even such a well studied set of diseases like cancer. Further complexity arises when pathological conditions emerge only when these driver mutations are interlinked in highly dependent combinations. Furthermore, these mutated proteins may only exert their influence under specific environmental conditions.
While it may be feasible for a few companies to be world leaders in knowledge about the human genome relative to many others, at this juncture this extra insight likely covers only a small fraction of the true complexity. The development of personalized medicine will require a lot more than knowing about the existence of troubling mutations in an individual's genome. Such information will probably be useful to less than a percent of the human population for many years to come. This is especially true when one also considers that the existing medical therapies for the most common and serious diseases are presently inadequate and typically treat the symptoms rather than the underlying causes.
In other words, genome sequencing might identify a lot of suspects, but proving which genes are guilty for a specific medical condition in a patient is totally another matter. Even if the culprit genes can be established, with 23,000 proteins encoded by the human genome and drugs currently available for only about 600, there are severe limitations on what can be done to rectify problematic proteins at this time.
Link to the original blog post.
At Kinexus Bioinformatics Corporation, we have recently completed a detailed meta-analysis of all of the reported mutations in over 3000 human genes that have been linked with cancer in one way or another. Much of the data was retrieved from the Sanger Institute's Catalogue of Somatic Mutations in Cancer (COSMIC) database, which contains over 217,000 mutations described for over 683,000 tumours. While there are some instances of well defined cancer-related proteins such as Ras isoforms and isocitrate dehydrogenase with clear hotspots of mutation, the vast majority feature mutations that are apparently randomly distributed throughout their sequences, and in most cases less than 1% of their amino acids have been observed to be altered with any cancer. In fact, for more than half of these cancer-associated proteins, either none or only 1 or 2 mutations have been discovered. In retrospect, this is not surprising, since gene mutation by radiation and chemical carcinogens is a random process, and a hallmark of advanced cancers is the accumulation of sporadic mutations due to damage to proteins that repair DNA.
With some 60 million single nucleotide polymorphisms likely to exist within the human genome, it will be some time before we can sort through all of the data and identify most of the bona fide pathologically relevant mutations in the human genome. This is certainly exemplified with the difficulty associated with assignment of driver and bystander mutations in even such a well studied set of diseases like cancer. Further complexity arises when pathological conditions emerge only when these driver mutations are interlinked in highly dependent combinations. Furthermore, these mutated proteins may only exert their influence under specific environmental conditions.
While it may be feasible for a few companies to be world leaders in knowledge about the human genome relative to many others, at this juncture this extra insight likely covers only a small fraction of the true complexity. The development of personalized medicine will require a lot more than knowing about the existence of troubling mutations in an individual's genome. Such information will probably be useful to less than a percent of the human population for many years to come. This is especially true when one also considers that the existing medical therapies for the most common and serious diseases are presently inadequate and typically treat the symptoms rather than the underlying causes.
In other words, genome sequencing might identify a lot of suspects, but proving which genes are guilty for a specific medical condition in a patient is totally another matter. Even if the culprit genes can be established, with 23,000 proteins encoded by the human genome and drugs currently available for only about 600, there are severe limitations on what can be done to rectify problematic proteins at this time.
Link to the original blog post.