From: Brian O'C. <boc...@uc...> - 2008-01-09 21:31:03
|
Funny that you're asking about this today. Jordan and I have been looking into this. We're having problems working with the image files Solexa provides. It's strange, they're 16-bit files (1000x1000x2 bytes/pixel) but looking at them using ImageMagic it seems like both the higher and lower bytes are identical e.g. it looks like they're essentially 8-bit images. We've been able to compress them using LZW (lossless) and reduce the file size to 1/5 the original size (which is awesome). The problem is running the solexa image file through ImageMagic causes changes in the file (it looks like the dynamic range is rescaled automatically by ImageMagic). When we do a round trip (tif -> lzw compressed tif -> tif), or even just a tif -> tif convertion, we see drastically different answers coming out of the pipeline (even when the dynamic range isn't reajusted!). I'm looking at the pipeline source code today to see if I can figure out what API they're using to read the images. But I'm hoping someone on the list or someone from Solexa can explain what's going on. --Brian Ben Berman wrote: > > what are you guys currently using to compress the images? I know you > mentioned this lossy one that's used for astronomy. If it's lossy, it > would be good to test the effect on the basecalling error when > uncompressed. We are writing a grant, so i want to decide if we're > going to keep the images or not. > > thanks, > ben. |