solution for incomplete key sequence 454 flx jr amplicon?

[brrt]

Does anyone have a solution for the following problem: We routinely perform 454 amplicon sequencing. In a recent 454 flx jr amplicon sequencing run, the first T flow has failed, prossibly due to an airbubble. We have obtained sufficient raw well reads from this run, and also the flow images look good. Since the first T flow is part of the key sequence, the 454 software cannot extract the data. As it is an amplicon sequencing run, the nucleotides following the key and MID sequence are known.
Is there a way to extract the data without the need of the T key sequence?

Bart

[Soulbee]

Why dont you try copying and pasting an image from another flow in the same run that has passed with a good brightness in place of the T image that has failed? With 454 GSFLX, the image files are numbered (ex. 00031.pif) and the numbering starts with the first image being 000.pif, so the image/flow number i see in gsBrowser will not be its correct file name... for example, if the failed T image was the flow number 30 in gsBrowser, it will have the file name 00029.pif. go find another image in the same run with desired brightness and copy that image file into the place of the T image file, but keep the original file name for T (00029.pif) basically changing the content of the file. NOTE that this is the naming system for FLX not the jr. so double check its file names.

after you have replaced the image, re run the data processing and the right key should be picked up. I suggest you make a copy of the whole sample folder then do the whole copy/pasting/re-processing on the duplicated folder, so the original file is intact.

[RCJK]

Would that actually work as the reads will not necessarily have the same location on the plate?
Would it be better to take an image from the same run that is known to have a T, such as if there is a T in their known MID or primer regions?

Edit: Sorry, I incorrectly read the beginning of Soulbee's post. I totally agree with what they said.

[brrt]

Thanks Soulbee. It is a very interesting solution. I had not realized that the individual images could be used to trick the image processing. I had one problem though, and that is that the first useful T is quite far in the run.. since first the MID sequence is analysed. I am trying now to see what happens if I just duplicate the C, G, or A flow image of the key sequence to replace the T. This may result in a structural over estimation or under estimation of the T in the rest of the sequence.. But since it is an amplicon run, for a large part I know what to expect. I will keep you posted on the results..

Bart

[Soulbee]

It shouldn't matter which image you replace it with because as far as the software is concerned, whatever image you put in place of the first T flow will in fact be the T flow. The software knows which of the 4 bases (T, C, A, or G) that particular wash was when that image was taken, so basically you are just changing the content for the "T file". The software will still think that the copied image you replaced is the T flow. Make sure you duplicate the whole folder of your run and re process with the duplicate. it is worth a try to force key recognition, but if it does not work, you'll probably have to repeat the run