Computational protocol: Conversion of cDNA differential display results (DDRT-PCR) into quantitative transcription profiles

Similar protocols

Protocol publication

[…] Detector signals from an ALFExpress II sequencer were converted into 16-bit TIFF images using ALF Sequence Analyzer 2.11 (Amersham Pharmacia Biotech AB, Uppsala, Sweden). Each chromatogram was exported using a low compression level (resampling factor of 2 was used for digital processing). Pseudogel images for digital processing were cropped to a length of 10,000 pixels by removing the lower part of the gel using Adobe Photoshop 6.0 (Adobe Systems Incorporated, San Jose, USA) and imported into the gel analysis software GelCompar 3.5 (Applied Maths, Kortrijk, Belgium). 10,000 pixel is the maximum gel length which can be processed by GelCompar. Gel image parts corresponding to fragments shorter than 50 bp were removed and lanes with artefacts were identified by manual inspection and discarded. A Cy5-labeled size standard (50-bp ladder) was defined as a standard marker lane on the first gel and the following gels were aligned to this marker.Band finding was performed automatically using a minimum profiling value of 0.5%, allowing only for bands with an elevation of at least 0.5% with respect to the surrounding background. Minimum area and shoulder sensitivity were set to zero. Further editing was performed manually for those band positions in which a band was found only in one of the two corresponding lanes (treatment/control). Bands were added or discarded according to the following criteria: when the densitometric curve revealed an elevation at the position in which a band was found in the corresponding lane, a new band was introduced and verified by checking whether its shape fitted to the densitometric profile. In case no elevation of the densitometry profile was found in a position in which a band was found in the corresponding lane and the corresponding band was weak (elevation barely over 0.5%), the band was discarded. Cases of strong bands in one lane corresponding to no band in the same position in the corresponding lane were very rare. Band matching was performed automatically for each two lanes corresponding to the same primer combination with position tolerance set to 1%. It was rarely necessary to adjust the matching manually, for example when several adjacent bands largely overlapped or when the algorithm failed to find obvious matches because of a slight retardation of lanes at the margin of the gel. Absolute peak area and calculated fragment size for each band were exported to Microsoft Excel (Microsoft, Redmond, USA).Normalization of densitometry values is crucial both for comparisons among profiles generated in the same experiment and among different experiments. Because the intensities of control and treatment lanes might differ due to differences in the efficiency of RNA extraction and RT-PCR and in the volumes loaded on the sequencer, band intensities (measured as peak areas) have to be adjusted by a normalization factor. This factor can be obtained as the mean of ratios of intensities of corresponding bands (treatment/control) for bands originating from transcripts that are unchanged by the treatment. A technical problem arises from the fact that a rigorous distinction between bands which are affected by the treatment and those which are not affected is only possible after the normalization, yet it is necessary to identify at least some unaffected bands for the calculation of the normalization factor. We solved the problem by assuming that only a minority of bands in each lane is affected by the treatment.To identify bands which were not affected by the treatment for the estimation of the normalization factor, we calculated "uncorrected induction factors" (UIF) for each pair of bands in corresponding lanes as the ratio of areas of peaks corresponding to matched bands. In the next step UIFs for all bands in a lane were sorted by values. Assuming that fewer than 25% of bands in a lane were induced and fewer than 25% suppressed by the treatment, the mean for all UIFs for bands between the first and third quartile was calculated. This value was used as a normalization factor to correct all absolute peak areas. Induction factors (IF) were now calculated as the ratio of normalized peak areas for matched bands in the treated sample and the control. Unaffected bands possessed induction factors close to 1.0, induced bands had IF > 1.0, suppressed bands had IF < 1.0. [...] Data were analysed using Microsoft Excel 7 (Microsoft, Redmond, USA) and graphs drafted with SigmaPlot 5.0 (SPSS, Chicago, USA). For results presented as Box-and-Whisker plots, the boxes include 50% of the ranked data, the whiskers show the 10th and 90th percentile, the outlier points mark the outliers defined as values above and below the 10th and 90th percentile. Means are expressed as arithmetic mean ± S.D. […]

Pipeline specifications

Software tools GelCompar, SigmaPlot, SPSS
Applications Miscellaneous, DNA fingerprinting
Organisms Brassica napus, Leptosphaeria maculans