PDF Documentation

Downloading FrameD and matrices

• Ralstonia solanacerum
• Sinorhizobium meliloti
• Sinorhizobium meliloti (pSymA)
• Bacteriophage P4
• Medicago truncatula (low GC)
• Medicago truncatula (high GC)
• Arabidopsis thaliana
• Bacillus subtilis
• Escherichia coli
• Mycobacterium tuberculosis
• Pyrococcus abyssi
• Encephalitozoon cuniculi

Filling in the form:

• Organism:  this indicates to FrameD to which organism the submitted sequence belongs. FrameD uses interpolated Markov Models (as in Glimmer) to identify regions that appear to be coding or not. If the organism is not present in the list, you may try to build a model of the coding regions by yourself: go the the markov model building page by clicking on "Learn a new one" just above the organism selection. Here, choose a name for your Markov model (name of the organism typically), and upload a FASTA file of CDS sequences of the organism. The CDS should go from the start to the STOP (excluded). If you leave the STOP at the end it will be automatically trimmed but if an inphase STOP exists beyond this last one the sequence will be rejected. FrameD uses interpolated Markov models which automatically adjust the order of the model to the amount of data but the larger the data set the better.
  If you later want to predict genes in noisi unfinished sequences, choose the "Build N resistant Markov model". If you don't do this, FrameD can still work on unfinished sequences using the "Use averaged Markov model (noisy sequence)" in its main web page.
• FASTA file: if you click to the button on the right (usually indicating "Browse..." but it is langage and browser dependent), you will see a file requester. Select the FASTA (or raw DNA) file that you want to submit. The filename will appear in the text window. Then click on any "Submit sequence" button to get a result. You may also prefer to just paste your sequence in the large text area below. You can either paste a FASTA format file or a raw DNA format sequence.
• BlastX file: if you click to the button on the right (usually indicating "Browse..." but it is langage and browser dependent), you will see a file requester. Select the NCBI BlastX output file that you want to submit. This file must be in the so-called "tabulated format" of NCBI-Blastx, obtained using the "-m8" flag of blastall The filename will appear in the text window. Then click on any "Submit sequence" button to get a result. You may also prefer to just paste your sequence in the large text area below. You can also paste this output below.
• Options:
• Compute mean expected prediction: FrameD always chooses a single "optimal" prediction. If you check this button, it will also compute a mean behavior that summarizes all the possible predictions that FrameD could do taking into account the more likely stronger. This is especially interesting for frameshifts. This affects the quantity of information presented in the graphical output.
• Increased sensitivity: by default, FrameD always decide if an ORF is coding or not by comparing the likeliness that it is coding (according to the usual statistical properties of the coding regions of the organim) with the likeliness that it is not (according to GC/AT frequency). Long ORFs that do not appear to be coding will not be detected as coding (which is very important for GC-rich organisms). However, is you want to take into account the length of ORFS in the prediction, you can check this button.  It is advisable to set the frameshift penalty to a maximum in this case.
• Frameshift penalty: indicates to FrameD how likely a frameshift is. The higher the penalty, the less likely the frameshifts. The default value of FrameD is around 12. If you enter a penalty above 1000, FrameD will forbid frameshift detection.
• Generate corrected seq.: when a  frameshift is detected (is part of the optimal prediction), FrameD will generate a sequence where frameshifts have been corrected. Detected inserted nucleotides will be removed and deleted nucleotides will be inserted back as 'N'.
• Generate translated seq.: for every gene predicted by FrameD that contains no predieted frameshift, the corresponding amino acid sequence is generated in a FASTA file. Each amino-acid sequence name is built from the DNA sequence name followed by the position of the gene in the original sequence. The phase of the gene follows as a comment.
• Eucaryote (ATG only): if you want to use FrameD on EST or cDNA, this will restrict start detection to ATG, activate the averaging propabilities model if needed, remove the RBS detection and change the prior probabilities to be more in phase with what eucaryotic genomes are.
• TGA is not a stop codon: for organism such that TGA is not a stop codon (Mycoplasma…).
• RBS pattern (16S side): by default, FrameD uses the standard E.Coli Shine Delgarno motif to detect RBS (the motif used is "attcctcc", a short version of the E.Coli motif atTCCTCCA, described in Ribosomes: structure, function and genetics, Steitz J.A.). If you want to use another motif (less than 10 nuc. long), just put it down here.
• Output:
• Text/Images/Both: select the type of output you want.
• Length per image: select the number of nucleotides presented per image. The default is 6000 (or the size of your sequence if it is shorter).
• Overlap between succ. images: select the amount of overlap between successives images (for long sequences). The default is heuristically determined.
• X-resolution (pixels): choose the horizontal resolution of the output. The larger the resolution, the finer the image but the larger the files. The default value is 900. A maximum of 1200 is enforced on the Web interface.
• Y-resolution (pixels): choose the vertical resolution of the output. The larger the resolution, the finer the image but the larger the files. The default value is 300. A maximum of 500 is enforced on the Web interface.
• Score smoothing window: all the statistics presented by FrameD on its graphical output are computed on a sliding window (coding/non coding curves, GC%, GC3%). By default the window used is 97 nucleotides wide. You can modify this here.
• Score normalisation: FrameD always compares the relative likelihood of being non coding or coding in the 6 phases. By default, the graphical output normalizes these 7 scores but you can ask for non normalization or to normalize independantly each coding phase w.r.t. the non coding hypothesis.
• Graphical elements: by default, frameD generates graphics with all possible information presented. If you have requested the mean behavior computation (Expectation), this may generates a very rich not so readable graphical output). You can select here the graphical elements that will appear in the output.

Reading the output:

• Text output:

• Text output: The text output simply presents the list of coding regions detected by FrameD with possible partial genes on the beginning and the end of the sequence and possible frameshifts.  For each coding region, the start, the end and the frame are indicated. A partial gene is indicated by a '>' (start missing) or a '>' (stop missing). Frameshifts are indicated by a vertical bar '|'.

  
  Loading parameters file...done
  Loading Markov model...(O8) (O8) (O8) done
  Loading sequence...1628 nuc. read
  GC Proportion = 67.7%
  Locating possible RBS...done
  Computing coding probabilities...done
  Gene parsing...done
  Optimal path length = -1971.218780 (0.18)
  
     Start    End  Ph.
  
       388   1422  1
      1495   1628  1   >
  
  

  Two genes are detected. The first gene start at 388 and finishes at 1422. The second gene is only partial (the stop is missing) and starts at 1495. The corresponding graphical output is:




• The horizontal axis is the sequence (here 1628 nuc. long).
• The vertical axis represents the 6 possible coding phases (+1,+2,+3 for the direct strand, -1, -2,-3 for the reverse) and  possible non coding (IG stands for intergenic) regions. Each of these is called a track in the sequel.
• On the 6 coding tracks, small red bars represent occurrences of STOP codons, blue bar represent START codons. The wider the blue bar, the better the start is (according to the presence of RBS, the type of the codon and how degenerated the codon is in the sequence eg. NNN is a possible start, but it is very degenerated).
• The fine black curve is a smoothed, normalized score of the coding/non coding appearance of the region.
• The light green curves indicates the GC% and the GC3% of the sequence. The GC% is indicated in the non coding track (IG) and is artificially amplified: the track contains variations from 25% to 75% GC. The GC3% indicates the percentage of GC bases in the 3rd position of the codons in this phase. It is not amplified (the track contains variation from 0% to 100%). The sequence presented above is extracted from Ralstonia solanacerum, a 68% GC rich genome. You can observe that the GC3% in the coding region is between 90 and 100%.
• The large red blocks indicate the prediction of FrameD. Thin red lines that connect two blocks indicate predicted frameshifts.
• Mean behavior: when the "expectation" computation is requested, FrameD will also compute a "mean prediction" and not only an optimal one. This mean prediction is represented by light gray lines. One can observe here, between 900 and 1000 that FrameD tends to predict that this region is coding on phase +2 rather than +1 (this would be possible by considering 2 frameshifts, one insertion around 900 and a deletion around 1000)
• 
  

• Changing the frameshift penalty: An example output for the very same sequence but with a frameshift penalty lowered to 6 is given below. Since the frameshift penalty has been lowered 2 frameshifts are detected in the optimal prediction. The frameshifts predicted are visible in the text output as vertical bars "|". Here one frameshift (an insertion) is detected between position 916 and 917 and a deletion is predicted between 1001 and 1002.
  
     Start    End  Ph.
  
       388    916  1   |
  |    917   1001  2   |
  |   1002   1422  1
      1495   1628  1   >
  
• We can ask for a sequence reconstruction by selecting the "-C" flag which is
  
  

Important note: one important restriction of FrameD is that it cannot detect two overlapping genes on opposite strands. When this rare situation appears, FrameD will usually detect a spurious frameshift to avoid overlapping and still detect both genes. This abnormal behavior is very easy to identify. It is expected that future versions of FrameD will be able to cope with such overlapping opposite strand genes.