Definitions

• the present invention relates to an image analysis method for measuring the signal on biochips.
• a biochip consists of a small solid support, generally a microscope slide, on which are grafted or synthesized in situ, at well-defined positions, DNA probes specific for cell gene sequences.
• the probes are grouped into a spot of a few square micrometers which contains approximately 10 9 identical molecules. All the spots thus form a two-dimensional matrix which can reach a complexity of 10 6 spots / cm 2 , as described in the document referenced [1] at the end of the description.
• the type of addressing of the probes on the support, as described in the document referenced [2] can vary and therefore impose a particular geometry:
• probes already synthesized by mechanical or electrochemical addressing are removed and placed on the slide by means of micropipettes or robotic needles.
• the support can be structured or unstructured.
• a biochip consisting of a matrix of wells having gold electrodes. Electropolymerization is carried out on each electrode between a normal pyroleum and a pyroleum on which a probe has been covalently attached.
• the probes are, for example, deposited by a robot which has a printhead with several needles, or rings.
• a robot with several rings, for example 4 makes it possible to take a small volume of the samples, the deposit itself being made by a small point which crosses the ring before coming into contact with the surface of the blade.
• a system robot of several, for example 48, hollow or solid needles makes it possible to take the samples and deposit them on the surface of the support.
• the geometry of the deposit depends on the geometry of the robot head, the type of plates used to store the probes, the size of the needles, the steps between each spot, etc. We generally obtain a set of spots grouped in several blocks separate, each block corresponding to a needle.
• FIG. 1 illustrates a deposit made with a head with four needles or rings with probe plates 10 and a biochip 11. After the reading of the slide by a scanner, the intensity (or the intensities if a mixture was initially co -hybridized) fluorescence is (are) quantified (s) at each spot.
• the acquisition of the images then consists in producing a digital image of the biochip after excitation in order to measure the intensities of the different fluorescence signals.
• the biochip is read by a scanner fitted with excitation lasers corresponding respectively to the absorption wavelengths of the fluorochromes used.
• the fluorescence intensity is measured by a photomultiplier.
• the quality and resolution of the images obtained depend on the scanner used. Different types of devices can be used, confocal and non-confocal scanners, CCD camera scanners ...: - a confocal scanner has a high quality lens system allowing to excite a small surface x, y over a small depth of field around the focal plane.
• the fluorescent emissions from this region are recovered at a point on the photomultiplier, this region corresponding to approximately one pixel of the image. Only the signal contained in the focal plane is read, everything outside this slice and this precise region does not reach the detector. Spurious signals, such as dust, are minimized if they are not in the focal plane of the lens. Otherwise, there is no mixing between the signals of two neighboring regions. The dust is therefore well insulated and can possibly be removed by further treatment. On the other hand, the acquisition plane being very thin, it may be necessary to carry out several passages on the slide to correct the reading errors of the photomultiplier and obtain less noisy images.
• - a non-confocal scanner does not allow such precise focusing.
• the recovered signal comes from a larger region and incorporates a large thickness of the blade.
• the intensity recovered at a specific point of the photomultiplier corresponds to an average in a certain neighborhood. There is therefore a signal dilution.
• the image is therefore very smooth and less contrasted, - a CCD camera scanner uses a high-precision camera for image acquisition.
• the quality of the image directly depends on the acquisition time. This reading step is very important.
• a good compromise between the contrast, the intensity of the background noise, and the resolution is essential to apprehend the difficulties of image analysis (location of spots, location of dust, segmentation of signals ...) and to optimize reliability results.
• the precision and resolution of the scanner are determining factors for the analysis of DNA microarrays.
• a digital image is obtained, for example, in a 16-bit Tiff format
• Tag-based Image File Format This format allows a good compromise between the size of the files to be stored and the quality of information, that is to say the number of intensity levels capable of being coded. It is therefore widely used to save images from many scanners.
• Tiff format is not specific to any system, it is easily manipulated and allows the addition of additional private information.
• each pixel is discretized on a scale of 65536 gray levels, from the lowest intensity (0) to the strongest (65635).
• a pixel is the smallest measurable area on the image. Its intensity represents the average fluorescence emitted by the surface of corresponding resolution on the biochip. It is the smallest hybridization unit measurable on an image of this type.
• the intensity is proportional to the number of cDNA molecules attached to the corresponding region of the slide.
• On a spot the density of the molecules is not necessarily homogeneous.
• the spots which are composed of a variable number of pixels of different intensities, can present a very heterogeneous signal. As illustrated in FIG. 2, an image is in the form of a large number of spots whose relative addresses are perfectly listed.
• spot Finding Problem The difficulty in automating a precise procedure for locating spots on the image of the biochip (Spot Finding Problem) comes from three main constraints:
• each spot is composed of a variable number of pixels of heterogeneous intensity. This is due, on the one hand, to the method of depositing the probes which, depending on the robots used, generates more or less circular spots and, on the other hand, to the amount of cDNA which hybridizes on the probes .
• the spots can have the shape of a ring, a disc or else consist of a set of small heterogeneous sub-regions.
• the software for extracting fluorescence signals of the known art are most often based on four steps:
• the localization of hybridization zones begins, in some software, with the semi-automatic placement of a grid. It is necessary to manually position each block of spots on the image. For each block, it is then necessary to provide the precise parameters of the two-dimensional matrix which constitutes it (number of lines, number of columns, steps between spots, spot size) in order to place a grid. By combining these geometric parameters and a protected analysis, the image is segmented into thumbnails inside which a more or less circular mask of variable size delimits the hybridization zone. This mask is centered on the maximum intensity. In software described in the document referenced [6], the spot location method is a little more automated and more reliable.
• the measurement of the fluorescence intensity of the spots is imposed by the software used and is not necessarily suitable for the type of deposit used.
• the object of the invention is therefore to solve the problems set out above, by proposing an image analysis method for measuring the reaction rate of the probes on a biochip.
• the invention relates to an image analysis method for measuring the signal on biochips organized in one or more blocks each comprising a large number of spots each composed of at least one probe, said method comprising the following steps: - a deposit is made with a constant step, the spots appearing in the form of a periodic and regular signal,
• this periodicity property is used to locate the blocks and the spots, this periodicity being sought on the x and y projections of the image, characterized in that it further comprises the following step:
• - degrading filters are used to increase the contrast of the image in order to position the spots using the periodic properties of the image, on the one hand, and non-degrading filters to minimize variations due to background noise and spot signal to amplify each spot.
• one uses averaged orthogonal projections with two axes x and y of the plane of the biochip making it possible to amplify the periodic signal, and one thus obtains two histograms H j and ⁇ f ⁇ .
• FFTD discrete one-dimensional Fourier transform
• a variable threshold can be applied to the two histograms R.
• a window of width ⁇ 0 and length 2 ⁇ 0 is used centered on the global cutting line, which makes it possible to locally explore the presence of signals that are too intense, which are therefore suspect, and possibly redefine the block limit around signals.
• the window can be extended by ⁇ 0/2 on one side of the limit then on the other, if there is an increase in the signal only for one of the two extensions of the window, the block boundary is moved by ⁇ 0 in the adjacent block for which there has been no signal increase, if there is no signal increase or if the signal has increased for the two displacements, this means either that it is a parasitic signal, or that the two blocks are very close, in these two cases we calculate the probability of belonging to the signal to the right and to the left of the cutting line, and we obtain P ( ⁇ ) for each half-projection with and without suspect signal, the gain obtained being equal to:
• the signal considered is assigned to it by displacement of the cutting line by ⁇ 0/2 in the adjacent block; if, for neither of the two blocks, the gain is not greater than 1, it is a parasitic signal, the cutting line is not moved; if the gain is greater than 1 for the two half-projections, the signal is in phase with the two half-projections, the blocks are therefore close and in phase and the signal is allocated to the block for which the gain is the greatest, by displacement of the section line of ⁇ 0/2 in the adjacent block.
• the binarized projection can be convolved by a sinusoidal function taken on ⁇ 0 such that:
• an image with a high resolution is produced, then the resolution of the image is artificially reduced by replacing n neighboring pixels. by the average of the n pixels which thus makes it possible to reduce the size of the image of n, and a homogeneity filter is applied to eliminate the pixel which exhibits the greatest heterogeneity with respect to the adjacent pixels.
• sorting the pixels within each spot makes it possible to identify the different types of pixels present: background noise, hybridization signal, spurious signal.
• an algorithm is applied
• Modified EM in which we narrow the classes and we introduce the notion of pixel not belonging to any class, by applying a threshold on the calculated probability that it belongs to one of the two or three main classes.
• An initialization step which consists of the construction of a fuzzy distribution table Ci (k) of the pixels
• An estimation step in which the new fuzzy distribution table is calculated
• the invention relates to an image analysis method which comprises the following steps:
• a step of automatic segmentation of the spot for. determine the pixels. representative of the reaction, and calculation of an evaluator of the spot signal by the mean, the median etc.
• a method is used based on the location of the local loss of the periodic signal, such as the power of the Fourier transform of the signal of the projection of the spots along the axes by lines or by columns.
• a first filter makes it possible to straighten the image, if necessary, by aligning the spots of a same block.
• a second filter can also amplify the signals coming specifically from the spots and level the other types of signals.
• the position of each block in the image is refined.
• Figure 1 illustrates a probe deposition made with a device of known art.
• Figure 2 illustrates a typical image of a chip.
• FIG. 3 illustrates different types of spots obtained.
• Figure 4 illustrates a Hi histogram
• FIG. 5 illustrates a diagram of
• Figure 7 illustrates an original image
• Figure 8 illustrates a cut image
• Figure 9 illustrates an image of a block.
• Figure 10 illustrates a Hi histogram.
• Figure 11 illustrates an alignment of the projections with a sinusoid.
• Figure 12 illustrates a Hi histogram
• Figure 13a illustrates a histogram after digital reconstruction.
• Figure 13b illustrates a binarized histogram.
• Figure 14a illustrates a projection and a sinusoidal function before adjustment.
• FIG. 14b illustrates a projection and a sinusoidal function after adjustment.
• FIG. 15 illustrates an example of a grid placed automatically on a block.
• FIG. 16 illustrates an example of spots obtained with mechanical deposition.
• Figure 17 illustrates the distribution of pixels in a spot.
• FIGS. 18a to 18d illustrate different images of the same spot acquired under four different conditions, FIGS. 18c and 18d being filtered under different conditions.
• FIG. 19 illustrates an example of a graph obtained by comparing, pixel to pixel, two images of the same spot taken at different wavelengths (different hybridization conditions).
• FIG. 20 illustrates an estimate of the different populations of pixels present in a spot with a conventional EM method.
• Figure 21a illustrates one. application of the modified EM method to find the pixel classes with a convergence evaluated with log-likelihood.
• Figure 21b illustrates an application of the modified EM method to find the pixel classes with convergence evaluated with RMS-log-likelihood.
• Figure 22a illustrates an image before filtering.
• FIG. 22b illustrates this image after degrading filtering.
• FIGS. 23a, 23b and 23c illustrate flowcharts which respectively describe a complete processing for a single image, a complete processing for several images of different hybridization conditions of the same slide with overlapping of the images, and a minimum processing for an image with use to implement the method of the invention.
• FIG. 24a represents an initial image of the chip
• FIG. 24b represents an image of the background noise obtained by extrapolating the background noise at the spot level.
• the method of the invention is an image analysis method for measuring the reaction rate of probes placed on a solid substrate forming a biochip organized in one or more blocks each comprising a large number of spots each composed of at least minus one probe.
• the images obtained during the reading phase are composed of several blocks of spots relatively well aligned and arranged in a constant pitch.
• the fixed geometry of the head of the robots used leads to the formation of blocks which can be approximated to homogeneous blocks x by y such as: 4 by 4 or 48 by 12, according to the number of needles available to this head. If the block is not homogeneous (blocks are missing at the end of the spot) the treatment will be done as a homogeneous block. Empty spot thumbnails will be eliminated later.
• the deposit being made with a constant step the spots appear in the form of a periodic and regular signal. Between each block of spots, the signal periodicity is lost.
• the method of the invention uses this periodicity property to locate the blocks and the spots.
• This periodicity is sought on the row and column projections, respectively in x and y, of the image.
• these projections cannot be used directly for all types of images. Delimitation of spot blocks
• the blocks are more or less well aligned in x and in y. Average projections orthogonal to the axes make it possible to amplify the main periodic signal, the singular artefactual signals being minimized. Two histograms R. and
• Ai, j intensity of a pixel with coordinates ( ⁇ , j), ni and n j : dimensions of the axes as a function of the height and width of the image.
• FFTD discrete one-dimensional Fourier transform
• the interval can possibly be adjusted to the frequency of the deposit if it is significantly different.
• a variable threshold is therefore applied to the two histograms H j and ⁇ .
• the value of this threshold is determined in the following manner. For a first threshold value S0, arbitrarily fixed very low, the ratio is calculated: ⁇ (3)
• FIG. 6 which illustrates the diagram of P ( ⁇ 0 ), at each of the calculation positions we obtain P ( ⁇ 0max ) along this window, for intensity values greater than the threshold that 1 'we have previously determined, with:
• P (C0 0 ) When arriving at the inter-block regions, P (C0 0 ) passes through a minimum which makes it possible to identify the limits of the blocks.
• the block limit is moved by ⁇ 0 in the adjacent block for which there has been no increase in the signal.
• the cutting line is displaced by ⁇ 0/2 on either side of the cutting line and P ( ⁇ ) is calculated (equation 3) for each half-projection obtained.
• Each movement alternately includes and then excludes the suspect signal at one of the half-projections. We therefore obtain P ( ⁇ ) for each half-projection with and without suspect signal. We can then calculate the gain obtained:
• the signal considered is assigned to it by displacement of the cutting line by ⁇ 0/2 in the adjacent block. If, for neither of the two blocks, the gain is not greater than 1, it is a spurious signal. The cutting line is not moved. If the gain is greater than 1 for the two half projections, the signal is in phase with the two half projections. The blocks are therefore close and in phase and the signal is assigned to the block for which the gain is greatest, by displacement of the cut line by ⁇ 0/2 in the adjacent block. Definitive membership in a block is determined by the cardinal or parity rules described later. The limits of each block can thus be defined. To avoid any encroachment in the neighboring block during displacements of the cutting limit (indeed it may be necessary to move the limits by a quantity lower or greater than the proposed values) the displacement can be carried out by searching for minima premises in the direction of travel.
• Figure 9 thus illustrates the image of a spot.
• it is very important to isolate the extinct spots that is to say the areas where probes have been deposited, but where no labeled cDNA of the analyzed mixture has hybridized.
• This absence of hybridization is important information for biologists.
• FIG. 10 thus illustrates a histogram R ..
• the projections are aligned using the least squares method with a sinusoid of equation:
• the average amplitude A of the projections is calculated as follows:
• the phase of the sinusoid is calculated by sliding along the profile of the histogram H A a sinusoidal pattern of amplitude A calculated over a period ⁇ 0 .
• the RMS Root Minimum Square
• Figure 11 illustrates the alignment of the projections with a sinusoid.
• FIG. 13a thus illustrates the histogram H. after digital reconstruction
• Figure 13b thus illustrates the H binarized histogram.
• the applied sinusoidal function is optimized such that:
• the reconstructed histogram makes it possible to find precisely the position of the spots presenting a strong contrast. By adding to it the information of the sinusoid which gives the theoretical positions, one can find all the spots presenting a contrast too weak to be detected.
• the binarized projection is convolved by a sinusoidal function taken on ⁇ 0 such that: .lambda.o
• H l ⁇ (A sin ( ⁇ time j) xHb l + J _ ⁇ 0I ) (il)
• Figure 14a illustrates a projection and a sinusoidal function before adjustment.
• Figure 14b illustrates the projection and the sinusoidal function after adjustment.
• the missing spots correspond to the maxima of the sinusoidal function (9) which have not been assigned to the convolution function.
• the position of the missing spots can therefore be deduced.
• the phase and spot size rules defined below are applied:
• a convolution function signal cannot be defined as a row or column of spots only if it is in phase with the maxima of the sinusoidal function (9),
• FIG. 15 illustrates an example of a grid placed automatically on a block.
• the blocks of a biochip are not necessarily the same size.
• the number of rows and columns depends on the deposit strategy chosen, the rules chosen being adaptable to the particular strategy.
• the homogeneity of the image obtained by DNA biochips depends on the type of deposit used: mechanical deposition or deposition by ink jet.
• FIG. 16 illustrates an example of spots obtained with mechanical deposition.
• FIG. 17 illustrates, in fact, the distribution of the pixels in the spot, The average of the intensity of the spot being calculated with the pixels greater than the average of the background noise pixels plus 1.5 times the standard deviation of the background noise .
• FIG. 18 illustrates different images of the same spot acquired under four different conditions, for example with a Packard scanner, and normalized at the same intensity; a) a single image with 10 ⁇ m resolution, fast passage, gain ⁇ 100 PMT 100
• FIGS. 18a and 18b the acquisition conditions are illustrated by FIGS. 18a and 18b.
• a variant consists in producing several images with low excitation and weak detection, which are combined into a single image.
• FIG. 18c there is a clear improvement, by a factor of 1.38, in the signal / noise ratio of the background noise or of the hybridization signals (see table 1 at the end of the description). This improvement is faster for the spot signal than for the background noise.
• a variant consists in producing a high resolution image, then in reducing the resolution of the image artificially by replacing n neighboring pixels by the average of the n pixels. This reduces the size of the image of n.
• a homogeneity filter can then be applied by eliminating the pixel which has the most great heterogeneity compared to adjacent pixels. Heterogeneity is estimated at more than 2 ⁇ (standard deviation) of the homogeneity estimator chosen.
• the homogeneity estimator chosen is the average of a window of 16 pixels centered on the four pixels to be averaged, ⁇ being calculated on the same window of 16.
• the four measurements carried out are independent. The probability that the same error will be made on all four pixels is very low. This gives an image with a noise intensity reduced by a factor of 1.8. In this case, the intensity of the background noise is reduced by a factor of 2.3 and the signal / noise ratio of the background noise or of the hybridization signals increases by a factor of 1.27. On the other hand, the average intensity and the variance of the hybridization pixels are only slightly modified, which results in a gain in the more modest signal / noise ratio by a factor of 1.12. Maintaining the hybridization signal and decreasing the average and the variance of the background noise by the proposed filter make it possible, as with accumulation, to decrease the intensity of the spot acceptance threshold. The problem of fluorescence quenching is thus eliminated.
• An alternative consists in using a filter which, from an image at 10 ⁇ m resolution of the DNA chip, is not or little degrading for the signal contained at the spot level, and minimizes the variance of the background noise .
• This filter increases the signal / noise ratio of the image spots and therefore increase the sensitivity of the chip.
• This filter is based on thumbnails delimiting the border between the position of the spots and that of the background noise of the image.
• thumbnails delimiting the border between the position of the spots and that of the background noise of the image.
• the background noise image thus obtained is subtracted from the original image. The resulting image is used to quantify the spots as described below.
• sorting the pixels within each spot makes it possible to identify the different types of pixel present: background noise, hybridization signal, parasitic signal.
• a modification of this method makes it possible to narrow the classes and to introduce the notion of pixel not belonging to any class, by applying a threshold on the calculated probability that it belongs to one of the two or three main classes.
• Figure 21a illustrates an application of the EM method thus modified, to find the pixel classes; convergence being estimated by the log likelihood with bg: estimated distribution of background noise, Real distribution: distribution of pixels of the hybridization signal, and offset: saturated pixel.
• Figure 21b illustrates an application of the modified EM method to find the pixel classes; convergence being estimated by RMS-log- likelihood with bg: estimated distribution of background noise, Real distribution: distribution of pixels in the spot, and offset: saturated pixel.
• the background noise is measured inside the spot, it is therefore no longer under or over rated, which eliminates the problem of negative spots.
• the method consists in classifying the pixels into three classes iteratively. To achieve such a classification, the maximization of the parameters of a mixture is used.
• the pixels are classified in two, then in three classes if necessary: background noise, hybridization signal and possibly saturated pixels.
• the first step consists in the construction of a fuzzy distribution table Ci 0 (k) of the pixels, where i is the intensity and k the class.
• This table illustrated in table 2 at the end of the description, is constructed in such a way that the value 1 is assigned for the background noise class and 0 for the class of the hybridization signal for pixels whose intensity is less than the background noise value + 2 sigma (taken as a reference outside the spot).
• the fuzzy distribution table Ci 0 (k) takes a value of 0 for the noise class background and 1 for the signal class, for pixels whose intensity is greater than the value of the background noise + 2 sigma (taken in reference outside the spot). At first, signals greater than 65500 are considered saturated and not taken into account.
• - ni is the number of pixels with the intensity level i
• Ci 0 (k) is the probability of assigning the intensity level i to class k and
• Ci 0 (k) is the probability of assigning the intensity level i to class k
• Steps 2 and 3 are repeated until the classification of the likelihood log defined as:
• RMS log-likelihood is such that:
• RMS l / R
• R being the correlation coefficient between the theoretical distribution calculated from the mixture and the actual distribution curve.
• the determination of the position of the spots is made all the more easily when there is contrast between background noise and spot signal. To increase this contrast, it is possible to use a degrading filter, which does not necessarily preserve the proportionality of the signals.
• the image thus filtered is only used to find the position of the spots. Quantification is done from the raw image, or from the image filtered by a non-degrading filter.
• - 500 and 65535 can be changed according to the amplitude of redistribution desired in the image.
• the threshold is chosen so as to maximize equation (3):
• a second "low pass" type filter assigns to each pixel (i, j) of the image the average of the signals of a window of (a + 1) side pixel centered on the pixel (i, j) such as:
• FIGS. 22a and 22b show the results presented in FIGS. 22a and 22b, FIG. 22a illustrating an image before filtering, FIG. 22b illustrating this image after filtering.
• FIG. 22a illustrating an image before filtering
• FIG. 22b illustrating this image after filtering.
• This amplification can be bidirectional by considering for example a cross of 2 ⁇ in height and 2 ⁇ in width, centered on ij such that:
• the amplifying function can be any periodic function taking into account the period of the phenomenon sought.
• the intensity of each pixel can be multiplied by a fraction of the power of the Fourier transform calculated in a window centered on the pixel considered.
• I s pot being the average intensity measured at the level of the background noise
• I bg being the average intensity of the background noise
• ⁇ spo t being the standard deviation of the signal
• ⁇ b g being the standard deviation of the background noise
• the histogram of the first m lines of the image is calculated as follows:
• each object found defines a class.
• the EM method merges or does not combine the classes of the various objects and makes it possible to define the class of pixels of the hybridization signal while excluding parasitic pixels originating from dust, for example.
• Figures 23a, 23b and 23c illustrate all the steps to be followed to quantify the spot, in three exemplary embodiments, respectively: a complete processing for a simple image, in which there are successively non-degrading filtering steps, filtering degrading, overall positioning of the blocks, refinement of the positioning of the blocks, delimitation of the spots, pixel segmentation of each spot and calculation of an intensity evaluator for each spot, - a complete processing for several images, of different hybridization conditions of the same support slide, with image superposition, which repeats the same steps as that of FIG. 23a with an image superposition step, - minimum processing for an image, in which there is no not the two preliminary filtering steps shown in Figure 23a.
• H LO complete processing for a simple image, in which there are successively non-degrading filtering steps, filtering degrading, overall positioning of the blocks, refinement of the positioning of the blocks, delimitation of the spots, pixel segmentation of each spot and calculation of an intensity evaluator for each spot
• Probe nucleic acid molecule whose sequence is complementary to a desired sequence. For DNA chips, they are immobilized on the support.
• DNA or Deoxyribonucleic Acid molecule supporting genetic information.
• MRNA Messenger Ribonucleic Acid. Molecular form in which the genetic message, encoded in genes, is transferred from the nucleus to the cytoplasm. This mRNA is used in the cytoplasm for the synthesis of a protein.
• CDNA or complementary DNA single-stranded DNA, which is complementary to an mRNA obtained by reverse transcription. Unlike DNA, it lacks introns (non-coding sequences). It can also be in double strand form, by copying the first strand with a DNA polymerase.
• Condition represents all of the RNA or cDNA extracted from cells in a particular state.
• Sequencing process used to determine the order (sequence) of amino acids of a protein or bases in nucleic acids (DNA and RNA).
• Retro-Transcription enzymatic reaction allowing the synthesis of a complementary DNA molecule from an mRNA molecule.
• PCR Polymerase Chain Reaction
• Molecular hybridization Effect of the pairing of two nucleic sequences. It is based on the principle of complementarity of nucleic bases, more particularly between DNA and the strand of RNA or cDNA. It can make it possible to highlight within a cell or a tissue, a nucleic acid sequence.
• Differential hybridization competitive hybridization of molecules from two different samples.
• Pixel Unit of intensity of the corresponding image, in the case of the scanner used in the laboratory, with an area of 5 or 10 microns depending on the resolution used.
• Sage “Serial Analysis of Gene Expression” (Nelculescu et al. 1995). A technique for quantitatively measuring the transcriptome of a cell.
• Genome All of the DNA of an individual or a species.
• Transcriptome Set of messenger RNAs transcribed from the genome into a cell at an instant t.
• Proteome Set of proteins in a cell at an instant t.

Landscapes

• Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
• Image Analysis (AREA)

Applications Claiming Priority (3)

Publications (1)

Family

ID=34134662

Family Applications (1)

Country Status (3)

Families Citing this family (1)

• 2003
  • 2003-06-18 EP EP03760739A patent/EP1514240A2/de not_active Withdrawn
  • 2003-06-18 AU AU2003258811A patent/AU2003258811A1/en not_active Abandoned
  • 2003-06-18 JP JP2004514943A patent/JP2005530167A/ja active Pending

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events