Hi Christian, On Mar 10, 10:53 am, cstratowa <[email protected] ingelheim.com> wrote: > Dear all, > > I am currently determining copynumber regions for tumor cell lines > using GLAD as model for the SNP6.0 array. (This takes about 13 hrs per > array and per cluster node)
Despite GLAD being a good segmentation algorithm, you may want to consider alternatives that are less computationally intensive and give equivalent results. CBS (package DNAcopy in bioconductor) runs in 1.5h for a SNP6.0 array. A fairly new but very promising program is Ultrasome (http://www.broad.mit.edu/ultrasome). It segments a SNP6.0 array in 2 seconds (plus 1 minute to read the array). It's a standalone program. > My problem is the interpretation of the results. > Here are some examples: > > 1. example: > "chromosome" "start" "stop" "mean" "count" "call" > 1 3674045 10539859 0.154251128612596 4517 > "neutral" > 1 10544961 10544961 1.57871123504559 1 "gain" > 1 10546623 10546623 -2.21390332082091 1 "loss" > 1 10548849 21751915 0.150400582947561 6872 > "neutral" > > This example has two regions where start=stop position (10544961, > 10546623). > How do you interpret this result? > How should regions with count=1 be interpreted? > Are these two regions considered to be outliers? SNP6.0 arrays are very noisy. You should be very careful when you consider any segment that has a size < 10 probes, and in any case, you should not believe any segment that has a size of one probe. In your case, I would just delete these segments. They are likely to be marked as "outliers" by GLAD. > 2. example: > "chromosome" "start" "stop" "mean" "count" "call" > 6 18253390 18253886 -1.33576824261601 2 "loss" > 6 18254264 29945167 -0.0986937482773974 7921 > "neutral" > 6 29945167 29945167 1.47894372023 1 "gain" > 6 29945294 30015940 0.225347608366043 42 "gain" > > Here you see two regions with different start positions but identical > stop positions (=29945167), where one region shows a gain but the > other region is neutral. > How do you interpret this result? Same problem, count is too small. > 3. example: > "chromosome" "start" "stop" "mean" "count" "call" > 8 15672606 15672655 -2.45103201175721 2 "loss" > 8 15672695 15672695 -4.68547262667068 1 "loss" > 8 15672695 15678118 -2.65573944523429 10 "loss" > 8 15678687 15678880 -2.01991964356680 3 "loss" > > In this case two regions have identical start positions (=15672695). > How can this be the case? > Could this be a bug, since it is assumed that two regions should not > overlap? > > 4. example: > "chromosome" "start" "stop" "mean" "count" "call" > 17 23813398 32995436 -0.272936124639797 5529 > "neutral" > 17 32999012 32999012 -2.26635873379834 1 "loss" > 17 32999012 32999012 -8.28862000630996 1 "loss" > 17 33000940 36676630 -0.295688057544771 1841 > "neutral" > > This example seems to be a combination of examples 2 and 3: > Here you have two identical regions with identical positions and > start=stop (=32999012) > How can two regions be identical but show a different mean (-2.266 vs > -8.288)? > Could this be a bug? Probably a bug, but again, this region should be discarded. > In summary my questions are: > How should I interpret the results? > Would it be better to eliminate these regions? > > Thank you in advance. > Best regards > Christian Nicolas --~--~---------~--~----~------------~-------~--~----~ When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example. You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/aroma-affymetrix?hl=en -~----------~----~----~----~------~----~------~--~---
