Hi Lei, Here are some links to the Table Browser help sections: http://genome.ucsc.edu/cgi-bin/hgTables?command=start#Help http://genome.ucsc.edu/goldenPath/help/hgTablesHelp.html
In particular, here is the link describing the intersection function: http://genome.ucsc.edu/goldenPath/help/hgTablesHelp.html#Intersection I can think of a few things that may be causing an intersection failure: 1) It is possible that no genes have yet been characterized that correspond to your peak data in the tracks you are intersecting against. This seems unlikely from your description, but it is possible that data was retracted from genbank (or whatever source was used) since the expression data was generated. It is also possible that the gene was unable to gain an alignment to genomic via BLAT for some reason or had a better alignment to a new location. 2) The intersection query parameters does not include the complete genomic for the target gene (UTR, intron, etc) so that a peak's coordinates miss the gene (some gene track's alignments represent coding regions only). There are options to adjust how much genomic is included for the comparison on the intersection form. 3) The intersection function only works for positional data (which is sounds like you are using). Perhaps the format has problems. If you want to send me some examples from your BED file, I will duplicate your query and provide feedback. Maybe send two queries you consider "pass" and two "fail". Let's start with mm8 (not the liftOver version) so we can rule out that process as a variable. Include for each: 1) Custom track header line(s) 2) Complete line from the BED file custom track input for a target gene 3) Expected "gene" that you thought would be identified via the intersection with this BED file line. Include the genomic coordinates as shown in the browser and name/accession. 4) Step-by-step what TB options you are using (everything, including the assembly version through to intersection options and output options) You can just send the examples to me, not the entire list. I will not repost your exact data to the list either, only our general solution, if appropriate. If you are able to figure this out on your own, just let me know. Jennifer Jackson UCSC Genome Bioinformatics Group Lei Ying wrote: > Hi, there: > This is Lei. I have problem with "intersection" function in "table > browser". > We have a custom track containing genome wide ChIP-Chip peak data in BED > format (mm8) and we try to find out which genes those peaks intersect. > Then we intersect our custom track with Knowgene table on Knowgene track > (mm8) and we found some genes with peaks are missing in the gene list. > The next thing we did is to liftover our BED format to mm9, and redo the > intersection. These genes still cannot be recognized after intersection. > We also tried to intersect our custom track with RefSeq table, and had > the same problem. > Do we miss some tricks during the procesee? If you could provide us a > detailed manual of how to use table browser, it will be a great help. > UCSC genome browser is something I cannot live without. THX for your > work. It's really cool. > Look forward to your reply. > Sincerely, > > > Lei Ying, MD, Ph.D > Postdoctoral Fellow > Department of Hematology > The Children's Hospital of Philadelphia > Philadelphia, PA, 19104 > _______________________________________________ > Genome maillist - [email protected] > http://www.soe.ucsc.edu/mailman/listinfo/genome > _______________________________________________ Genome maillist - [email protected] http://www.soe.ucsc.edu/mailman/listinfo/genome
