@-1. TX 0 @General @-2. T 0 @Screen control @-2. X 0 @Screen @-3. TX 0 @Set parameters @-4. TX 0 @Comparison @0. TX -1 @SIP This is program for comparing and aligning nucleic acid or protein sequences. It can produce optimal alignments using a dynamic programming algorithm, and has several ways of producing "dot matrix" diagrams. The following analyses (preceded by their option numbers) are included: The program is used on a simple graphics terminal ie a keyboard with a screen on which points and lines can be drawn. The user works at the terminal and produces plots for various combinations of values for the span length and minimum scores. However large or small a region the user elects to compare the program expands or contracts the diagram so that the plot always fills the screen. This allows the user to gain an overall impression or to "home-in" on particular regions and examine them in more detail. Having found a region that looks interesting the user can determine its coordinates in terms of sequence positions by use of a crosshair facility. The program has two statistical options to help the user choose score levels for plotting and to assess the significance of any similarity found. It can produce a cumulative histogram of observed scores for the current span length and region and it can calculate the "double matching probability" of McLachlan. The "double matching probability" is the probability of finding particular scores given two infinitely long sequences of the composition of those being compared, with the current span length and score matrix. By using these options the user can choose to plot all the matches for which the score exceeds a given significance level (such as 1%), using either empirical or theoretical probability values. Generally it is best to begin at a low level to avoid an overcrowded diagram. If the user finds that the two sequences do contain stretches of homology he will often want to align the sequences by inserting padding characters at deletion points. The program has a selection of options for this purpose: it contains an alignment routine; it can display on the screen the two sequences, one above the other, with asterisks marking identities, it has inbuilt editing functions and can save the aligned sequences on disk files. The basic principle of dot matrices was first described by Gibbs and McIntyre and involves producing a diagram that contains a representation of all the matches between a pair of sequences. This diagram is then scanned by eye and the human ability to recognise patterns used to detect any similarities that might be present. The diagram consists of a two dimensional plot in which the x axis represents one sequence (A) and the y axis the other (B). Every point (i,j) on the plane x,y is assigned a score which corresponds to the level of similarity between sequence characters A(i) and B(j). In the simplest use of the method a score of 1 could be assigned to every point (i,j) where A(i) = B(j), and a score of 0 to every other point. If a plot of the points in the plane was made in which all scores of 1 were marked with a dot and all those of 0 left blank then regions of identity would appear as diagonal lines. With the comparison displayed in this form the human eye is very good at detecting regions of homology even if they are imperfect. The effects of mismatches, insertions or deletions can be seen: matches interrupted by insertions or deletions will appear as parallel diagonals, and matches interrupted by the odd mismatching pair of characters will appear as broken collinear diagonal lines. This diagram is a very useful representation but simply placing a dot for every identity is of limited value for the following reasons. For nucleic acid sequences around 25% of the plot will contain points and it will often be very difficult to distinguish significant homologies from chance matches. For proteins many significant alignments of sequences contain almost no identities but are formed from chemically and structurally similar amino acids so that simply looking for identity would be insufficient. What is required is to first find those points that correspond to fairly strong local similarities and then to use the diagram of these points so that the human eye can be used to look for larger scale homologies. The program uses a number of different algorithms to calculate the score for each point and the user defines a minimum score so that only those points in the diagram for which the score is at least this value will be marked with a dot. The first scoring method finds the longest uninterrupted sections of perfect identity i.e. those that contain no mismatches, insertions or deletions. Generally this method, termed "the identities algorithm" is of little value, but runs very quickly. The second method looks for sections where a proportion of the characters in the sequence are similar, again allowing no insertions or deletions. For a thorough analysis this method, termed "the proportional algorithm", is the best. The original method, of this type was first described by McLachlan and involves calculating a score for each position in the matrix by summing points found when looking forwards and backwards along a diagonal line of a given length. This length, called the span, must be an odd number so that the dot marking matches can be precisely placed at its centre. The algorithm does not simply look for identity but uses a score matrix that contains scores for every possible pair of characters. For comparing amino acid sequences we usually use the score matrix shown below which was calculated by adding 10 (to make every term >0) to each term of the relatedness odds matrix MDM78 of Dayhoff. This matrix MDM78 was calculated by looking at accepted point mutations in 71 families of closely related proteins and, of those tested by Dayhoff, was found to be the most powerful score matrix for finding distant relationships between amino acid sequences. AMINO ACID SCORE MATRIX ----------------------- C S T P A G N D E Q B Z H R K M I L V F Y W - X ? C 22 10 8 7 8 7 6 5 5 5 5 5 7 6 5 5 8 4 8 6 10 2 10 10 10 10 S 10 12 11 11 11 11 11 10 10 9 10 10 9 10 10 8 9 7 9 7 7 8 10 10 10 10 T 8 11 13 10 11 10 10 10 10 9 10 10 9 9 10 9 10 8 10 7 7 5 10 10 10 10 P 7 11 10 16 11 9 9 9 9 10 9 10 10 10 9 8 8 7 9 5 5 4 10 10 10 10 A 8 11 11 11 12 11 10 10 10 10 10 10 9 8 9 9 9 8 10 6 7 4 10 10 10 10 G 7 11 10 9 11 15 10 11 10 9 10 10 8 7 8 7 7 6 9 5 5 3 10 10 10 10 N 6 11 10 9 10 10 12 12 11 11 12 11 12 10 11 8 8 7 8 6 8 6 10 10 10 10 D 5 10 10 9 10 11 12 14 13 12 13 12 11 9 10 7 8 6 8 4 6 3 10 10 10 10 E 5 10 10 9 10 10 11 13 14 12 12 13 11 9 10 8 8 7 8 5 6 3 10 10 10 10 Q 5 9 9 10 10 9 11 12 12 14 11 13 13 11 11 9 8 8 8 5 6 5 10 10 10 10 B 5 10 10 9 10 10 12 13 12 11 13 11 11 10 10 8 8 6 8 5 7 4 10 10 10 10 Z 5 10 10 10 10 10 11 12 13 13 11 14 12 10 10 8 8 8 8 5 6 4 10 10 10 10 H 7 9 9 10 9 8 12 11 11 13 11 12 16 12 10 8 8 8 8 8 10 7 10 10 10 10 R 6 10 9 10 8 7 10 9 9 11 10 10 12 16 13 10 8 7 8 6 6 12 10 10 10 10 K 5 10 10 9 9 8 11 10 10 11 10 10 10 13 15 10 8 7 8 5 6 7 10 10 10 10 M 5 8 9 8 9 7 8 7 8 9 8 8 8 10 10 16 12 14 12 10 8 6 10 10 10 10 I 8 9 10 8 9 7 8 8 8 8 8 8 8 8 8 12 15 12 14 11 9 5 10 10 10 10 L 4 7 8 7 8 6 7 6 7 8 6 8 8 7 7 14 12 16 12 12 9 8 10 10 10 10 V 8 9 10 9 10 9 8 8 8 8 8 8 8 8 8 12 14 12 14 9 8 4 10 10 10 10 F 6 7 7 5 6 5 6 4 5 5 5 5 8 6 5 10 11 12 9 19 17 10 10 10 10 10 Y 10 7 7 5 7 5 8 6 6 6 7 6 10 6 6 8 9 9 8 17 20 10 10 10 10 10 W 2 8 5 4 4 3 6 3 3 5 4 4 7 12 7 6 5 8 4 10 10 27 10 10 10 10 - 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 X 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 ? 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 It is also possible to use other matrices, including an identity matrix for proteins. For nucleic acids we usually use the matrix shown below. DNA SCORE MATRIX A C G T X A 1 0 0 0 0 C 0 1 0 0 0 G 0 0 1 0 0 T 0 0 0 1 0 X 0 0 0 0 0 Plotting dots at the centres of spans that reach the cutoff leads to a persistence effect that, to some extent, can be mitigated by a variation on the method. If, for example, all the high scoring amino acids are clustered at the left end of a particular diagonal segment, dots will continue to be plotted to their right until the span score drops below the cutoff. Instead of plotting a single point for each span that reaches the cutoff score, the variant method plots points for all the identities that lie in spans that reach the cutoff. Obviously the persistence effect can be more pronounced for long spans and low cutoff scores, but note that the variant method will not plot anything if there are no identities present, and so similar regions could be missed! A further variant, useful for comparing a sequence against itself, ignores the main diagonal. The third comparison method called "quick scan" is really a combination of the first two, and is similar to the FASTP program of Lipman and Pearson, but produces a dot matrix diagram. The algorithm is as follows. The dot matrix positions are found for all words of some minimum length (obviously length 1 is most sensitive) that are common to both sequences. Imagine a diagonal line running from corner to corner of the diagram, at right angles to the diagonals in the dotmatrix, The scores for the common words (according to the current score matrix, e.g. MDM78) are accummulated at the appropriate positions on that imaginary line, hence producing a histogram. The histogram is analysed to find its mean and standard deviation. The diagonals that lie above some cutoff score (defined in standard deviation units), are rescanned using the proportional algorithm, and a diagram produced. The method is very fast, and is also employed by the library comparison program. The dynamic programming alignment algorithm contained in the program is based on that of Miller and Myers (). It guarantees to produce alignments with the optimum score given a score matrix, a gap start penalty, and a gap extension penalty. That is, starting a gap costs a fixed penalty (IG) and each residue added to the gap incurs a further penalty (IH) so that for each gap of length K residues the penalty is IG + k*IH. Gaps at the ends of sequences incur no penalty. It is very useful to have the dot matrix methods and the alignment routine together in the same program because it allows users to produce a dot matrix diagram to help select which regions of the sequence they wish to align. Selection is made by use of the crosshair. First the crosshair is positioned at the bottom left hand end of the segment to be aligned. The crosshair function is quit and immediately selected again, the crosshair positioned at the top right of the segment, and the crosshair function quit. When the alignment routine is selected the segment will be aligned. The alignment can replace the original segment of the sequence. By repeated plotting of dot matrices, followed by alignment, very long sequences can easily be aligned. @1. TX 0 @Help This option gives online help. The user should select option numbers and the current documentation will be given. The following analyses (preceded by their option numbers) are included: ? = Help ! = Quit 3 = read a new sequence 4 = define active region 5 = list the sequence 6 = list a text file 7 = direct output to disk 8 = write active sequence to disk 9 = edit the sequences 10 = clear graphics screen 11 = clear text screen 12 = draw a ruler 13 = use cross hair 14 = reposition plots 15 = label diagram 16 = display a map 17 = apply identities algorithm 18 = apply proportional algorithm 19 = list matching spans 20 = set span length 21 = set proportional score 22 = set identities score 23 = calculate expected scores 24 = calculate observed scores 25 = show current parameter settings 26 = quick scan 27 = draw a / 28 = align the sequences 29 = complement the sequences 30 = switch main diagonal 31 = switch identities 32 = change score matrix @2. TX 0 @Quit This function stops the program. @3. TX 1 @Read a new sequence This option allows users to read in new sequences, browse through annotations, or search sequence libraries for keywords. Sequences can be read from "personal" sequence files or from sequence libraries. These are referred to as the sequence "source". Personal files can be stored in several formats: Staden, PIR, EMBL, GENBANK and GCG. At LMB we use "Staden" format for sequencing and all the libraries are stored in their original formats. Note, however, that libraries such as EMBL or GenBank that are divided into several files (eg GenBank has 13 separate files) are indexed as a whole. This means that users do not need to know which file contains an entry, only which library. When the user selects to read in a sequence the program first asks for the sequence "source". If the user selects "personal" the program will ask for the format (Staden, PIR, EMBL, GENBANK or GCG), and then for the name of the file. For PIR format the user will also be required to know the entry name of the sequence as the file can contain several. For the other formats only a single entry is expected. The file will be read, its length and composition will be displayed and the option left. If the user selects "library" as the sequence source the program will display a list of available libraries. The programs are capable of handling all current libraries but which ones are available will vary from site to site. At LMB we have several libraries and also weekly updates of data gathered between releases. The program will ask users to select a library and then give a list of options: X 1 Get a sequence 2 Get annotations 3 Get entrynames from accession numbers 4 Search titles for keywords 5 Search text index for keywords If get a sequence or get annotations is selected users will be asked to type the entry name. The option will be left when a sequence is selected or ! is typed. The composition and length will be displayed. The text index contains all words from feature tables, reference titles, definition lines, keywords lists and comments, so the text index search is most useful. It is also the fastest. Up to 5 words can be searched for at once. The words should be typed separated by spaces, for example ? Keywords=P53 mouse murine tumo will search for all entries that contain words starting with p53, mouse, murine and tumo. Only the unique entries that contain ALL words will be listed. Before listing the matching entries the program will show the number of 'hits' for each word and ring the bell. Escape is possible at this point, or after each screenfull of entries. In addition to the entry names the text search displays the primary accession number, the sequence length and up to 80 characters of description. (The search of 'titles' is now redundant because the full text index contains all the title words and the search is much faster. It will probably be removed from the program.) All searches are independent of case. Where possible the program will offer default entry names. Typical dialogue follows. Select sequence source X 1 Personal file 2 Sequence library ? Selection (1-2) (1) = Select sequence file format X 1 Staden 2 EMBL 3 GenBank 4 PIR 5 GCG ? Selection (1-5) (1) = ? Sequence file name=M13MP7.SEQ Contig title removed Sequence length= 7238 Sequence composition T C A G - 2405. 1539. 1765. 1527. 2. 33.2% 21.3% 24.4% 21.1% 0.0% . . . Select sequence source X 1 Personal file 2 Sequence library ? Selection (1-2) (1) =2 Select a library X 1 EMBL 29 nucleotide library Dec 91 2 SWISSPROT 20 protein library Nov 91 3 PIR 31 protein library Dec 91 4 NRL3D 58 From Brookhaven protein library Dec 91 5 GenBank ? Selection (1-5) (1) = Library is in EMBL format with indexes Select a task X 1 Get a sequence 2 Get annotations 3 Get entry names from accession numbers 4 Search titles for keywords 5 Search text index for keywords ? Selection (1-5) (1) =5 Search for keywords ? Keywords=P53 mouse P53 hits 68 MOUSE hits 8180 MMANT01 X00875 536 Murine gene fragment for cellular tumour antigen MMANT02 X00876 83 Murine gene fragment for cellular tumour antigen MMANT03 X00877 21 Murine gene fragment for cellular tumour antigen MMANT04 X00878 261 Murine gene fragment for cellular tumour antigen MMANT05 X00879 184 Murine gene fragment for cellular tumour antigen MMANT06 X00880 113 Murine gene fragment for cellular tumour antigen MMANT07 X00881 110 Murine gene fragment for cellular tumour antigen MMANT08 X00882 137 Murine gene fragment for cellular tumour antigen MMANT09 X00883 74 Murine gene fragment for cellular tumour antigen MMANT10 X00884 107 Murine gene for cellular tumour antigen p53 (exon MMANT11 X00885 562 Murine p53 gene 3' region with exon 11 MMANTP53 M26862 536 Mouse tumor antigen p53 gene, 5' end. MMLYN M64608 2044 Mouse lyn protein mRNA, complete cds. MMP53 X00741 1377 Mouse mRNA for transformation associated protein MMP53A M13872 1285 Mouse p53 mRNA, complete cds, clone pcD53. MMP53B M13873 1241 Mouse p53 mRNA, complete cds, clone p53-m11. MMP53C M13874 1322 Mouse p53 mRNA, complete cds, clone p53-m8. MMP53G1 X01235 554 Mouse genomic DNA for 5' region of cellular tumou MMP53IN4 X60470 729 M.musculus p53 gene for p53 protein, intron 4 MMP53P X01236 2132 Mouse pseudogene for cellular tumour antigen p53 MMP53R X01237 1773 Mouse mRNA for cellular tumour antigen p53 MMRSB2P5 M64597 196 Mouse B2 repeat in the 3' flank of protein 53 (p5 22 different entries found Select a task X 1 Get a sequence 2 Get annotations 3 Get entry names from accession numbers 4 Search titles for keywords 5 Search text index for keywords ? Selection (1-5) (1) =4 Search for keywords ? Keywords=alpha Searching for alpha AAGHA 623 a.anguilla mrna for glycoprotein hormone alpha subunit precu AAMALI 3338 a.aegypti mali gene encoding alpha 1-4 glucosidase, complete AAMALIA 1659 a.aegypti maltase-like i (mali) gene encoding alpha-1,4-gluc AAMALIB 1832 a.aegypti maltase-like i (mali) mrna encoding alpha-1,4-gluc ACA13GT 371 alouatta caraya alpha-1,3gt gene, 3' flank. ADHBADA1 102 duck alpha-d-globin gene, exon 1. ADHBADA2 1145 duck alpha-a-globin gene and 5' flank ADHBADWP 513 duck (white pekin) alpha ii (minor) globin mrna, complete co AEACOXABC 5279 a.eutrophus protein x (acox), acetoin:dcpip oxidoreductase-a AGA13GT 371 ateles geoffroyi alpha-1,3gt gene, 3' flank. AGAAAGFP 282 c.tetragonoloba alpha-amylase/alpha-galactosidase fusion pro AGAABL 138 b.subtilis alpha-amylase signal peptide gene e.coli beta-lac AGAFAMYA 57 synthetic b.stearothermophilus alpha amylase/s.cerevisiae ma AGAFAMYB 57 synthetic b.stearothermophilus alpha amylase/s.cerevisiae ma AGAFAMYC 57 synthetic b.stearothermophilus alpha amylase/s.cerevisiae ma AGAFCOXA 98 synthetic alpha-factor/cox iv fusion gene signal peptide. AGAGABA 7876 synthetic gossypium hirsutum (cotton) alpha globulin a and b AGAMYLS 120 synthetic alpha-amylase gene, 5' end. AGANPS 95 synthetic gene (jcnf-1) encoding alpha-factor pro-region/han ! Select a task X 1 Get a sequence 2 Get annotations 3 Get entry names from accession numbers 4 Search titles for keywords 5 Search text index for keywords ? Selection (1-5) (1) =3 ? Accession number=v00636 Entry name LAMBDA Select a task X 1 Get a sequence 2 Get annotations 3 Get entry names from accession numbers 4 Search titles for keywords 5 Search text index for keywords ? Selection (1-5) (1) =2 Default Entry name=LAMBDA ? Entry name= ID LAMBDA standard; DNA; PHG; 48502 BP. XX AC V00636; J02459; M17233; X00906; XX DT 03-JUL-1991 (Rel. 28, Last updated, Version 3) DT 09-JUN-1982 (Rel. 1, Created) XX DE Genome of the bacteriophage lambda (Styloviridae). XX KW circular; coat protein; DNA binding protein; genome; KW origin of replication. XX OS Bacteriophage lambda OC Viridae; ds-DNA nonenveloped viruses; Siphoviridae. XX RN [1] RP 1-48502 RA Sanger F., Coulson A.R., Hong G.F., Hill D.F., Petersen G.B.; RT "Nucleotide sequence of bacteriophage lambda DNA"; RL J. Mol. Biol. 162:729-773(1982). XX ! Select a task X 1 Get a sequence 2 Get annotations 3 Get entry names from accession numbers 4 Search titles for keywords 5 Search text index for keywords ? Selection (1-5) (1) = Default Entry name=LAMBDA ? Entry name= DE Genome of the bacteriophage lambda (Styloviridae). Sequence length 48502 Sequence composition T C A G - 11988. 11360. 12336. 12818. 0. 24.7% 23.4% 25.4% 26.4% 0.0% @4. TX 1 @Define active region For its analytic functions the program always works on a region of the sequence called the active region. When a new sequence is read into the program the active region is automatically set to start at the beginning of the sequence and go up to the maximum allowed size of active region the program can handle. The positions are shown on the screen. On most machines this will be to the end of the sequence. This option allows the user define a different region. @5. TX 1 @List a sequence The sequence can be listed with line lengths from 10 to 120 in multiples of 10. The output looks like: 87 97 107 117 127 137 KVKCTGRILE VPVGRGLLGR VVNTLGAPID GKGPLDHDGF SAVEAIAPGV IERQSVDQPV ** * **** *** * ** * * ** * ** * DVKDLEHPIE VPVGKATLGR IMNVLGEPVD MKGEIGEEER WAIHRAAPSY EELSNSQELL 68 78 88 98 108 118 147 157 167 177 187 197 QTGYKAVDSM IPIGRGQREL IIGDRQTGKT ALAIDAIINQ RDSGIKCIYV AIGQ ** * * * * * * *** * * * ETGIKVIDLM CPFAKGGKVG LFGGAGVGKT VNMMELIRNI AIEHSGYSVF AGVG 128 138 148 158 168 178 @6. TX 1 @List a text file Allows the user to have a text file displayed on the screen. It will appear one page at a time. @7. TX 1 @Direct output to disk Used to direct output that would normally appear on the screen to a file. Select redirection of either text or graphics, and supply the name of the file that the output should be written to. The results from the next options selected will not appear on the screen but will be written to the file. When option 7 is selected again the file will be closed and output will again appear on the screen. @8. TX 1 @Write active region to disk This option allows users to write the current active sequence to a disk file in Staden format. @9. TX 1 @Edit the sequences This function allows the user to insert or delete parts of either sequence to help align them. The inserted characters are dashes. @10. TX 2 @Clear graphics Clears the screen of both text and graphics. @11. TX 2 @Clear text Clears only text from the screen. @12. TX 2 @Draw a ruler This option allows the user to draw a ruler or scale along the axes of the screen to help identify the coordinates of points of interest. The user can define the position of the first sequence element to be marked (for example if the active region is 1501 to 8000, the user might wish to mark every 1000th element starting at either 1501 or 2000 - it depends if the user wishes to treat the active region as an independent unit with its own numbering starting at its left edge, or as part of the whole sequence). The user can also define the separation of the ticks on the scale and their height. If required the labelling routine can be used to add numbers to the ticks. To escape type ! @13. TX 2 @Use cross hair This function puts a steerable cross on the screen that can be used to find the coordinates of points in the sequence. The user can move the cross around using the directional keys; when he hits the space bar the program will write out the coordinates of the cross in sequence units and the option will be exited. If instead, the user hits a , the position will be displayed but the cross will remain on the screen. If a letter s is hit the sequences around the cross hair are displayed as a short alignment (as shown below) and the cross remains on the screen. 97 107 VPVGRGLLGR VVNTLGAPID **** *** * ** * * VPVGKATLGR IMNVLGEPVD 78 88 If a letter m is hit the sequences around the cross hair are displayed in the form of a matrix (as shown below) and the cross remains on the screen. VPVGKATLGRIMNVLGEPVD D...................DD I..........I.........I P.P...............P..P A.....A..............A G...G....G......G....G L.......L......L.....L T......T.............T N............N.......N VV.V..........V....V.V VV.V..........V....V.V R.........R..........R G...G....G......G....G L.......L......L.....L L.......L......L.....L G...G....G......G....G R.........R..........R G...G....G......G....G VV.V..........V....V.V P.P...............P..P VV.V..........V....V.V VPVGKATLGRIMNVLGEPVD The function is also used prior to "align sequences" in order to delineate the region to be aligned. The crosshair is positioned at the bottom left of the region, the crosshair option quit. Then the crosshair option is selected again, and the crosshair moved to the top right of the region to be aligned. @14. TX 2 @Reposition plots The position of the plots is defined relative to a users drawing board which has size 1-10,000 in x and 1-10,000 in y. Plots are drawn in a window defined by x0,y0 and xlength,ylength. Where x0,y0 is the position of the bottom left hand corner of the window, and xlength is the width of the window and ylength the height of the window. --------------------------------------------------------- 10,000 1 1 1 -------------------------------------- ^ 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ylength 1 1 1 1 1 1 1 1 1 1 1 1 -------------------------------------- v 1 1 x0,y0^ 1 1 <---------------xlength--------------> 1 --------------------------------------------------------- 1 1 10,000 All values are in drawing board units (i.e. 1-10,000, 1-10,000). The default window positions are read from a file "DIAGMARG" when the program is started. Users can have their own file if required. This option allows users to change window positions whilst running the program. If the user types only carriage return for any value it will remain unchanged. The cross-hair can be used to choose suitable heights. @15. TX 2 @Label a diagram This routine allows users to label any diagrams they have produced. They are asked to type in a label. When the user types carriage return to finish typing the label the cross-hair appears on the screen. The user can position it anywhere on the screen. If the user types R (for right justify) the label will be written on the diagram with its right end at the cross-hair position. If the user types L (for left justify) the label will be written with its left end at the cross hair position. The cross-hair will then immediately reappear. The user may put the same label on another part of the diagram as before or if he hits the space bar he will be asked if he wishes to type in another label. @16. TX 2 @Display a map NOT AVAILABLE. This draws a map of any sequence features selected by the user. These features may be protein coding regions (CDS), tRNA genes (TRNA), promoter positions (PRM), etc. Users may define their own feature table key names. The coordinates must be stored in a file in the format of an EMBL feature table. @17. TX 4 @Apply identities algorithm The identities algorithm finds runs of identical characters in the sequence. Its main value is speed, being 100's of times faster than the proportional algorithm. It is of course not very sensitive, and should only be used for a quick scan. The cutoff score is the minimum number of consecutive matching characters. All runs of identical characters that are at least as long as the cutoff score will produce a dot on the screen. See also quick scan. Typical dialogue follows. ? Menu or option number=d17 ? Identity score (1-20) (2) =3 Working missing graphics @18. TX 4 @Apply proportional algorithm This method, generally the most useful, was first described by McLachlan and involves calculating a score for each position in the matrix by summing points found when looking forwards and backwards along a diagonal line of a given length. This length, called the span, must be an odd number. The algorithm does not simply look for identity but uses a score matrix that contains scores for every possible pair of characters. At each point that a threshold score is achieved the program marks the screen in one of two ways. It will either place a single dot at the position corresponding to the centre of the matching span, or it will plot a dot for each identical residue within each matching span. Alternatively, the "list matching spans" option will list the segments that match. For comparing amino acid sequences we usually use the score matrix shown below which was calculated by adding 10 (to make every term >0) to each term of the relatedness odds matrix MDM78 of Dayhoff. This matrix MDM78 was calculated by looking at accepted point mutations in 71 families of closely related proteins and, of those tested by Dayhoff, was found to be the most powerful score matrix for finding distant relationships between amino acid sequences. AMINO ACID SCORE MATRIX ----------------------- C S T P A G N D E Q B Z H R K M I L V F Y W - X ? C 22 10 8 7 8 7 6 5 5 5 5 5 7 6 5 5 8 4 8 6 10 2 10 10 10 10 S 10 12 11 11 11 11 11 10 10 9 10 10 9 10 10 8 9 7 9 7 7 8 10 10 10 10 T 8 11 13 10 11 10 10 10 10 9 10 10 9 9 10 9 10 8 10 7 7 5 10 10 10 10 P 7 11 10 16 11 9 9 9 9 10 9 10 10 10 9 8 8 7 9 5 5 4 10 10 10 10 A 8 11 11 11 12 11 10 10 10 10 10 10 9 8 9 9 9 8 10 6 7 4 10 10 10 10 G 7 11 10 9 11 15 10 11 10 9 10 10 8 7 8 7 7 6 9 5 5 3 10 10 10 10 N 6 11 10 9 10 10 12 12 11 11 12 11 12 10 11 8 8 7 8 6 8 6 10 10 10 10 D 5 10 10 9 10 11 12 14 13 12 13 12 11 9 10 7 8 6 8 4 6 3 10 10 10 10 E 5 10 10 9 10 10 11 13 14 12 12 13 11 9 10 8 8 7 8 5 6 3 10 10 10 10 Q 5 9 9 10 10 9 11 12 12 14 11 13 13 11 11 9 8 8 8 5 6 5 10 10 10 10 B 5 10 10 9 10 10 12 13 12 11 13 11 11 10 10 8 8 6 8 5 7 4 10 10 10 10 Z 5 10 10 10 10 10 11 12 13 13 11 14 12 10 10 8 8 8 8 5 6 4 10 10 10 10 H 7 9 9 10 9 8 12 11 11 13 11 12 16 12 10 8 8 8 8 8 10 7 10 10 10 10 R 6 10 9 10 8 7 10 9 9 11 10 10 12 16 13 10 8 7 8 6 6 12 10 10 10 10 K 5 10 10 9 9 8 11 10 10 11 10 10 10 13 15 10 8 7 8 5 6 7 10 10 10 10 M 5 8 9 8 9 7 8 7 8 9 8 8 8 10 10 16 12 14 12 10 8 6 10 10 10 10 I 8 9 10 8 9 7 8 8 8 8 8 8 8 8 8 12 15 12 14 11 9 5 10 10 10 10 L 4 7 8 7 8 6 7 6 7 8 6 8 8 7 7 14 12 16 12 12 9 8 10 10 10 10 V 8 9 10 9 10 9 8 8 8 8 8 8 8 8 8 12 14 12 14 9 8 4 10 10 10 10 F 6 7 7 5 6 5 6 4 5 5 5 5 8 6 5 10 11 12 9 19 17 10 10 10 10 10 Y 10 7 7 5 7 5 8 6 6 6 7 6 10 6 6 8 9 9 8 17 20 10 10 10 10 10 W 2 8 5 4 4 3 6 3 3 5 4 4 7 12 7 6 5 8 4 10 10 27 10 10 10 10 - 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 X 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 ? 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 One alternative for proteins is to use an identity matrix. For comparing nucleic acids we usually use the matrix shown below. DNA SCORE MATRIX A C G T X A 1 0 0 0 0 C 0 1 0 0 0 G 0 0 1 0 0 T 0 0 0 1 0 X 0 0 0 0 0 See option 32 for how to change the score matrices. When a sequence is compared against itselt to look for repeats it is possible to use the proportional algorithm in a mode such that the main diagonal is not shown. See option 30. Typical dialogue follows. ? Menu or option number=d18 ? Odd span length (1-401) (11) = ? Proportional score (1-297) (132) = Working missing graphics @19. TX 4 @List matching spans This option applies the proportional algorithm using the current span and cut-off score, but instead of drawing a dot matrix it lists all the matching spans. When a sequence is compared against itselt to look for repeats it is possible to use this algorithm in a mode such that the main diagonal is not listed. See option 30. Typical dialogue follows. ? Menu or option number=d19 ? Odd span length (1-401) (11) = ? Proportional score (1-297) (132) =148 List matching spans Working 76 IEVPVGKATLG LEVPVGRGLLG 95 77 EVPVGKATLGR EVPVGRGLLGR 96 78 VPVGKATLGRI VPVGRGLLGRV 97 79 PVGKATLGRIM PVGRGLLGRVV 98 85 LGRIMNVLGEP LGRVVNTLGAP 104 86 GRIMNVLGEPV GRVVNTLGAPI 105 87 RIMNVLGEPVD RVVNTLGAPID 106 @20. TX 3 @Set span length The proportional algorithm calculates a score for each position in the matrix by summing the points found when looking forwards and backwards along a diagonal line of a given length. This length, called the span, should be an odd number so that the score for any point is correctly positioned at the centre of the span. This option allows the user to define the span length. It should be noted that short spans can produce noisy diagrams, but are less affected by insertions and deletions than are long spans. However long spans can detect more distant relationships. Long spans can suffer from a persistence problem by plotting dots when all the "signal" is to one side of the spans central position. To help avoid this, the option that plots the position of all matching residues within a matching span, can be tried. This is most useful if an identity matrix is being used. @21. TX 3 @Set proportional score The proportional algorithm calculates a score for each position in the matrix by summing the scores for the individual amino acids found when looking forwards and backwards along a diagonal line of a given length. All points at which the proportional score is achieved will produce a dot on the diagram. (The same score is used for the 'LIST MATCHING SPANS' option.) Before chosing a score the user can apply the routine that will calculate the expected score, or can calculate a histogram of observed scores. It is best to start with a high score to avoid an overcrowded diagram. @22. TX 3 @Set identities score The identities algorithm is of limited value as it only finds runs of matching characters, however it has the virtue of being very fast. This option allows the user to set the minimum length of run that will produce a dot on the screen. @23. TX 3 @Calculate expected scores This function calculates the "double matching probability" of McLachlan. The "double matching probability" is the probability of finding particular scores given two infinitely long sequences of the composition of those being compared, with the current span length and score matrix. By using this option the user can choose to plot all the matches for which the score exceeds a given significance level (such as 1%). Generally it is best to begin at a low level to avoid an overcrowded diagram. When the calculation of the expected scores is finished the program offers the user 3 ways of examining the results: "Show probability for a score" allows the user to type in a score and the program responds with the probability of achieving that level of score. "Show score for a probability" allows the user to type in a probability value and the program types the score that corresponds to that level of probability. "List scores and probabilities" is the command to list out the scores and their corresponding probabilities. The user is asked to supply a further parameter, the "number of steps between scores", and the program only lists every stepsize point. e.g a stepsize of 5 will get every 5th score listed. Typical dialogue follows. ? Menu or option number=d23 ? Odd span length (1-401) (11) = ? Proportional score (1-297) (132) = Working Average score= 103.18557 RMS deviation= 7.85276 X 1 Show probability for a score 2 Show score for a probability 3 List scores and probabilities ? 0,1,2,3 = ? Show probability for score (1-165) (134) =160 Probability of score 160 is 0.0000000008 X 1 Show probability for a score 2 Show score for a probability 3 List scores and probabilities ? 0,1,2,3 =2 ? Show score for probability (0.0000000001-1.) (0.00001) =0.0000001 Score for probability 0.0000001000 is 153 1 Show probability for a score X 2 Show score for a probability 3 List scores and probabilities ? 0,1,2,3 =3 ? Number of steps between scores (1-10) (5) = 0 0.10000E+01 100 0.67232E+00 200 0.18977E-20 5 0.10000E+01 105 0.42119E+00 205 0.42561E-22 10 0.10000E+01 110 0.20671E+00 210 0.87767E-24 15 0.10000E+01 115 0.78860E-01 215 0.16651E-25 20 0.10000E+01 120 0.23515E-01 220 0.27300E-27 25 0.10000E+01 125 0.55406E-02 225 0.00000E+00 30 0.10000E+01 130 0.10443E-02 230 0.00000E+00 35 0.10000E+01 135 0.15935E-03 235 0.00000E+00 40 0.10000E+01 140 0.19906E-04 240 0.00000E+00 45 0.10000E+01 145 0.20569E-05 245 0.00000E+00 50 0.10000E+01 150 0.17758E-06 250 0.00000E+00 55 0.10000E+01 155 0.12938E-07 255 0.00000E+00 60 0.10000E+01 160 0.80360E-09 260 0.00000E+00 65 0.10000E+01 165 0.43009E-10 265 0.00000E+00 70 0.10000E+01 170 0.20049E-11 270 0.00000E+00 75 0.99997E+00 175 0.82263E-13 275 0.00000E+00 80 0.99949E+00 180 0.29998E-14 280 0.00000E+00 85 0.99448E+00 185 0.98050E-16 285 0.00000E+00 90 0.96543E+00 190 0.28934E-17 290 0.00000E+00 95 0.86836E+00 195 0.77556E-19 295 0.00000E+00 1 Show probability for a score 2 Show score for a probability X 3 List scores and probabilities ? 0,1,2,3 =! @24. TX 3 @Calculate observed scores This option applies the proportional algorithm to the currently active sequence but instead of producing a dot matrix it calculates a histogram of observed scores. The speed of this calculation of course depends on the size of the active regions, but when it is completed the program offers the user 3 ways of examining the results: "Show percentage for score" allows the user to type in a score and the program responds with the percentage of points that achieve this value. "Show percentage for score" allows the user to type in a percentage and the program responds with the corresponding score. Values of this score and above are only achieved by the given percentage of points. "List scores and percentages" is the command to list out the scores and the percentage of points achieving them. Typical dialogue follows. ? Menu or option number=24 Working Maximum observed score is 152 X 1 Show percentage reaching a score 2 Show score for a percentage 3 List scores and percentages ? 0,1,2,3 = ? Show percentage for score (1-152) (114) =144 Percentage of points with score 144 is 0.005486297 X 1 Show percentage reaching a score 2 Show score for a percentage 3 List scores and percentages ? 0,1,2,3 =2 ? Show score for percentage (0.00001-1.) (0.001) =0.01 Score for percentage 0.010000000 is 143 1 Show percentage reaching a score X 2 Show score for a percentage 3 List scores and percentages ? 0,1,2,3 = ? Show score for percentage (0.00001-1.) (0.001) =1. Score for percentage 1.000000000 is 124 1 Show percentage reaching a score X 2 Show score for a percentage 3 List scores and percentages ? 0,1,2,3 =3 ? Number of steps between scores (1-10) (5) =1 73 236953 0.10000E+03 74 236951 0.99999E+02 75 236951 0.99999E+02 76 236950 0.99998E+02 77 236945 0.99996E+02 78 236942 0.99995E+02 79 236929 0.99989E+02 80 236900 0.99977E+02 missing data here 130 384 0.16206E+00 131 307 0.12956E+00 132 239 0.10086E+00 133 180 0.75964E-01 134 134 0.56551E-01 135 103 0.43468E-01 136 78 0.32918E-01 137 67 0.28276E-01 138 46 0.19413E-01 139 40 0.16881E-01 140 33 0.13927E-01 141 29 0.12239E-01 142 24 0.10129E-01 143 19 0.80184E-02 144 13 0.54863E-02 145 10 0.42202E-02 146 8 0.33762E-02 147 7 0.29542E-02 148 7 0.29542E-02 149 6 0.25321E-02 150 5 0.21101E-02 151 3 0.12661E-02 152 3 0.12661E-02 1 Show percentage reaching a score 2 Show score for a percentage X 3 List scores and percentages ? 0,1,2,3 =! @25. TX 3 @Show current parameter settings This function lists the names of the current sequences, their total lengths, the start and end points of the active sequence and the current values of span and cut-off scores. It also shows if the main diagonal will be shown, or if the proportional algorithm will mark all identities in matching spans. Typical dialogue follows. ? Menu or option number=25 Horizontal sequence ALPHA.PRT Positions 1 TO 514 Vertical sequence BETA.PRT Positions 1 TO 461 Span length= 11 Scores Proportional= 132 Identities= 3 Identites off Main diagonal shown @27. TX 2 @Draw a / This option simply draws a diagonal line from the bottom left of the diagram to the top right. it can be an aid when trying to align the sequences. @26. TX 4 @Quick scan The algorithm is as follows. The dot matrix positions are found for all words of some minimum length (obviously length 1 is most sensitive) that are common to both sequences. Imagine a diagonal line running from corner to corner of the diagram, at right angles to the diagonals in the dotmatrix, The scores for the common words (according to the current score matrix, e.g. MDM78) are accummulated at the appropriate positions on that imaginary line, hence producing a histogram. The histogram is analysed to find its mean and standard deviation. The diagonals that lie above some cutoff score (defined in standard deviation units), are rescanned using the proportional algorithm, and a diagram produced. The method is very fast, and is also employed by the library comparison program. Typical dialogue follows. ? Menu or option number=d26 ? Identity score (1-20) (3) = ? Odd span length (1-401) (11) = ? Proportional score (1-297) (132) = ? Number of sd above mean (0.00-10.00) (5.00) = missing graphics SIPL the library searching version of SIP This program compares a probe sequence against a library of sequences using the quick scan algorithm, sorts the matches into descending order of score, and produces optimal alignments of the best scores using the Myers and Miller method. It is very rapid. Use of lists of entry names SIPL has the ability to restrict searches to subsets of the libraries. This does not require sublibraries to be created but instead is achieved by using files containing a list of the entry names of sequences. The user may choose to search only those entries on the list or, alternatively to search all but those on the list (i.e. in the latter case the list contains the names of those to be excluded). The programs can search libraries that have indexes and those that do not. If a list of names for inclusion is used, then the search will be faster if the index is present. In all other circumstances the whole library will be read. The list must be in library order except when it is used to include entries, and an index is available. The list must contain each entry name on a separate line, with the name starting in column 1 of the line. ie there must be no spaces at the start of the line. The list of entry names can be produced by the keyword searches of nip, pip, sip, etc as long as the listings produced have a space character separating the entry name from the entry description. This will depend on how well the library reformatting programs work. For example swissprot entry names tend to run into the beginning of the descriptions, but other libraries are generally OK. @28. TX 4 @Align sequences This function will produce an optimal alignment of two segments of the sequence. The dynamic programming alignment algorithm is based on that of Miller and Myers (). It guarantees to produce alignments with the optimum score given a score matrix, a gap start penalty, and a gap extension penalty. That is, starting a gap costs a fixed penalty (F) and each residue added to the gap incurs a further penalty (E) so that for each gap of length K residues the penalty is F + K*E. Gaps at the ends of sequences incur no penalty. The routine can only handle segments of sequence of maximum length 5000 residues. When the sequences are read in the alignment segment will be set to the first 5000 residues. A different segment can be selected by prefixing the option number by the letter D, in which case the cross hair can be used to identify the two ends. The cross hair will appear. First position the crosshair at the bottom left of the segment and type a character other than s or m or ",". When the crosshair reappears, position it a the top right of the segment, and type a keyboard character. The aligned sequences will replace the active sequence if the user confirms "keep alignment". By alternate use of the plotting and alignment routines it is possible to rapidly produce an alignment of quite long sequences. Typical dialogue follows. 28 = Align sequences ? Menu or option number=d28 Define the region to align using the cross-hair. First identify the bottom left position and exit the cross-hair routine. Then the top right. (Bell rings, type return, cross hair appears) ? Penalty for starting a gap (1-100) (10) = ? Penalty for each residue in gap (1-100) (10) = Aligning region 1 to 461 with region 1 to 514 1 11 21 31 41 51 MA--TGKIVQ VIGA------ VVDVEFPQDA VPRVYDALEV QNG------N ERLVL----- * * * ** * * * * * MQLNSTEISE LIKQRIAQFN VVSEAHNEGT IVSVSDGVIR IHGLADCMQG EMISLPGNRY 1 11 21 31 41 51 61 71 81 91 101 111 EVQQQLGGGI VRTIAMGSSD GLRRGLDVKD LEHPIEVPVG KATLGRIMNV LGEPVDMKGE * * ** * * ** ***** *** * ** * * ** AIALNLERDS VGAVVMGPYA DLAEGMKVKC TGRILEVPVG RGLLGRVVNT LGAPIDGKGP 61 71 81 91 101 111 121 131 141 151 161 171 IGEEERWAIH RAAPSYEELS NSQELLETGI KVIDLMCPFA KGGKVGLFGG AGVGKTVNMM * ** * ** * * * * * * *** LDHDGFSAVE AIAPGVIERQ SVDQPVQTGY KAVDSMIPIG RGQRELIIGD RQTGKTALAI 121 131 141 151 161 171 181 191 201 211 221 231 ELIRNIAIEH SGYS-VFAGV GERTREGNDF YHEMTDSNVI DKVSLVYGQM NEPPGNRLRV * * ** * * * DAI--INQRD SGIKCIYVAI GQKASTISNV VRKLEEHGAL ANTIVVVATA SESAALQYLA 181 191 201 211 221 231 241 251 261 271 281 291 ALTGLTMAEK FRDEGRDVLL FVDNIYRYTL AGTEVSALLG RMPSAVGYQP TLAEEMGVLQ * * *** * * * * * * ** * * * RMPVALMGEY FRDRGEDALI IYDDLSKQAV AYRQISLLLR RPPGREAFPG DVFYLHSRLL 241 251 261 271 281 291 301 311 321 331 341 351 ERITST---- ---------- -KTGSITSVQ AVYVPADDLT DPSPATTFAH LDATVVLSRQ ** **** * * * * * * ERAARVNAEY VEAFTKGEVK GKTGSLTALP IIETQAGDVS AFVPTNVISI TDGQIFLETN 301 311 321 331 341 351 361 371 381 391 401 411 IASLGIYPAV DPLDSTSRQL DPLVVGQEHY DTAR----GV QSILQRYQEL KDIIAILGMD ** *** * * ** * * * * * ** LFNAGIRPAV NPGISVSR-- ---VGGAAQT KIMKKLSGGI RTALAQYREL AAFSQFAS-- 361 371 381 391 401 411 421 431 441 451 461 471 ELSEEDKLVV ARARKIQRFL SQ----PFFV AE----VFTG SPGKYVSLKD --TIRGFKGI * * * * * * * * * * * * DLDDATRKQL DHGQKVTELL KQKQYAPMSV AQQSLVLFAA ERG-YLADVE LSKIGSFEAA 421 431 441 451 461 471 481 491 501 511 521 MEG--EYDHL P-EQAFYMVG SIEEAVE--- --------KA KKL* ** * * * * * LLAYVDRDHA PLMQEINQTG GYNDEIEGKL KGILDSFKAT QSW* 481 491 501 511 521 Conservation 22.5% Number of padding characters inserted 63 and 10 ? (y/n) (y) Keep alignment n @29. TX 1 @Complement the sequences This function allows users to reverse and complement nucleic acid sequences. @30. TX 3 @Switch main diagonal If a sequence is being compared against itself to look for repeats it is sometimes convenient if the main diagonal is not included in the comparison. This function allows users to set a switch that determines whether or not to include the main diagonal for all the comparison methods. If the switch is set, and the active regions for both sequences have the same start position, then the main diagonal will not be compared. @31. TX 3 @Switch identities This function allows a switch to be set or unset. The switch determines which of two forms of plot will be produced by the proportional algorithm. One form of output (the original method) plots a dot at the centre of each span that reaches the threshold score; whereas the other form will plot dots for all matching residues that lie within spans that reach the threshold. @32. TX 3 @change score matrix This option allows users to select their own score matrix for use with the proportional algorithm. The choices are: 1 = MDM78 2 = identity 3 = your own matrix MDM78 is the standard matrix that is used for proteins and an identity matrix is the default matrix for nucleic acids. However an identity matrix is also useful for protein comparisons. "Your own matrix" allows users to apply any other matrix, as long as the matrix file is in the same format as MDM78. For comparisons of DNA it might be useful to try one that gave say 3 for exact matches and 1 for R-R or Y-Y, else=0. @33. TX 3 @Set number of sd's for Quickscan The quickscan algorithm is as follows. The dot matrix positions are found for all words of some minimum length (obviously length 1 is most sensitive) that are common to both sequences. Imagine a diagonal line running from corner to corner of the diagram, at right angles to the diagonals in the dotmatrix, The scores for the common words (according to the current score matrix, e.g. MDM78) are accummulated at the appropriate positions on that imaginary line, hence producing a histogram. The histogram is analysed to find its mean and standard deviation. The diagonals that lie above some cutoff score (defined in standard deviation units), are rescanned using the proportional algorithm, and a diagram produced. This option allows the number of sd's to be set. @34. TX 3 @Set gap penalities The alignment function will produce an optimal alignment of two segments of the sequence. The dynamic programming alignment algorithm is based on that of Miller and Myers (). It guarantees to produce alignments with the optimum score given a score matrix, a gap start penalty, and a gap extension penalty. That is, starting a gap costs a fixed penalty (F) and each residue added to the gap incurs a further penalty (E) so that for each gap of length K residues the penalty is F + K*E. Gaps at the ends of sequences incur no penalty. This option allows the gap penalties to be set. @ end of help