PHRED Documentation ------------------- 1. Introduction. Phred reads DNA sequencer trace data, calls bases, assigns quality values to the bases, and writes the base calls and quality values to output files. Phred can read trace data from SCF, ABI model 373 and 377 DNA sequencer chromatogram, and MegaBACE ESD chromatograms files, automatically detecting the file format, and whether the chromat file was compressed by gzip or UNIX compress. After calling bases, phred writes the sequences to files in either FASTA format, the format suitable for XBAP, PHD format, or the SCF format. Quality values for the bases are written to FASTA format files or PHD files, which can be used by the phrap sequence assembly program in order to increase the accuracy of the assembled sequence. Significant differences in this release New - add X86_GCC_LINUX definition. Defining this (in the supplied Makefile) when compiling phred on x86 Linux machines using the GCC compiler, causes phred to call bases and assign quality values that are identical to those called and assigned when phred runs on IEEE conforming UNIX machines. See the INSTALL and Makefiles in this distribution and the Linux system file /usr/include/fpu_control.h for additional information. - add LI-COR quality value lookup tables for chromatograms created by the LI-COR sequencing machine. See the section below entitled LI-COR Data. Modified - set `source' string, which is written in the output SCF files, based on machine type inferred from primer ID and phredpar.dat rather than based on the chromatogram type. The possible `source' values are "ABI 373A or 377" "MegaBACE" "ABI 3700" "LI-COR 4000" "unknown" - add `-trim_fasta', `-trim_scf', and `-trim_phd' options, which cause phred to write trimmed sequence and quality values to FASTA, SCF, and PHD files, respectively. Also added `-trim_out', which causes phred to write trimmed sequence and quality values to all output files (FASTA, SCF, and PHD) except .poly files. This is the same as using `-trim_fasta', `-trim_scf', and `-trim_phd' in combination. See the section on `Sequence Trimming' below. NOTE: `-trim_phd' affects the values in the `TRIM' field of the PHD file comment block. - describe trimming in greater detail in the section `Sequence Trimming'. - add protection against buffer overflows in code 2. Acknowledgements. Phred benefits from ideas developed by LaDeana Hillier, Mike Wendl, Dave Ficenec, Tim Gleeson, Alan Blanchard, and Richard Mott. 3. Algorithms. Phred uses simple Fourier methods to examine the four base traces in the region surrounding each point in the data set in order to predict a series of evenly spaced predicted locations. That is, it determines where the peaks would be centered if there were no compressions, dropouts, or other factors shifting the peaks from their "true" locations. Next phred examines each trace to find the centers of the actual, or observed, peaks and the areas of these peaks relative to their neighbors. The peaks are detected independently along each of the four traces so many peaks overlap. A dynamic programming algorithm is used to match the observed peaks detected in the second step with the predicted peak locations found in the first step. Phred evaluates the trace surrounding each called base using four or five quality value parameters to quantify the trace quality. It uses a quality value lookup table to assign the corresponding quality value. The quality value is related to the base call error probability by the formula QV = - 10 * log_10( P_e ) where P_e is the probability that the base call is an error. Phred uses data from a chemistry parameter file called 'phredpar.dat' in order to identify dye primer data. For dye primer data, phred identifies loop/stem sequence motifs that tend to result in CC and GG merged peak compressions. It reduces the quality values of potential merged peaks and splits those peaks that have certain trace characteristics indicative of merged CC and GG peaks. In addition, the chemistry and dye information are passed to phrap. 4. Building and installing. The INSTALL file describes the steps for building and installing phred. Copy the phred parameter file, called 'phredpar.dat', to a directory that is accessible by phred users and set the environment variable 'PHRED_PARAMETER_FILE' to the full path name of the file. For example, if you copy 'phredpar.dat' to '/usr/local/etc/PhredPar' and you are using the C shell then issue the command % setenv PHRED_PARAMETER_FILE /usr/local/etc/PhredPar/phredpar.dat It is most convenient to set the environment variable in the system- wide shell startup (cshrc or equivalent) file. You can rename the phred parameter file but the PHRED_PARAMETER_FILE environment variable must reflect the new name. With Windows NT you give the command % set PHRED_PARAMETER_FILE=\usr\local\etc\PhredPar\phredpar.dat in the DOS command window in which you will run phred. Note: if you compile phred on a SUN Solaris OS using the BSD C compiler in the directory `/usr/ucb', you will find that the `-id' command line option fails (phred reports that it cannot read files, and it prints the name of each file it fails to read; however, the name it prints lacks the first few characters of the true name of the file). If this occurs, recompile phred using either the optional C compiler in the directory /opt/SUNWspro/bin or the GNU C compiler. 5. Running phred. Phred uses command line options to control input, processing, and output. The command line options are delimited by a dash, "-". The command line options are Input Options ------------- -id Read and process files in . -if Read and process files listed in the file . Each line in must specify a valid path to a single input file. -zd Location of compression program. If -zd is omitted, phred uses the current path to search for the compression program. -zt Directory where chromat is uncompressed. If -zd is omitted, phred uses /usr/tmp. When phred processes a compressed file, it uncompresses the chromat into this temporary directory before it reads the file. It subsequently deletes the uncompressed file in the temporary directory. Processing Options ------------------ -nocall Disable phred base calling and set the current sequence to the ABI base calls that are read from the input file. By default, the current sequence is set to the phred base calls. This affects the base trimming and output options. -trim Perform sequence trimming on the current sequence. Bases are trimmed from the start and end of the sequence on the basis of trace quality. In addition, specifies a base sequence that is used to trim bases off the start of the current sequence. You can specify a NULL enzyme sequence using empty double quotes, "". See the note below on the effect of using the trim option. -trim_alt Perform sequence trimming on the current sequence. Bases are trimmed from the start and end of the sequence on the basis of trace quality. Specifically, for each base, the phred error probability is subtracted from the default value of 0.05 (or the value set using the `-trim_cutoff' option), and the resulting values are summed to find the maximum scoring subsequence. Furthermore, the subsequence must have a minimum number of bases. In addition, specifies a base sequence that is used to trim bases off the start of the current sequence. You can specify a NULL enzyme sequence using empty double quotes, "". -trim_cutoff Set trimming error probability for the `-trim_alt' option and the trimming points written in the phd files. The default value is 0.05. -trim_fasta Trim sequences written to sequence and quality value FASTA files. Set trimming information in the FASTA headers to reflect the high quality of the sequence, and append the string `trimmed' to the header. -trim_scf Trim sequence, quality values, and base locations written to SCF file. Append the string `trimmed' to the comments. -trim_phd Trim sequence, quality values, and base locations written to PHD files. Also set the first and last high quality base locations specified in the `TRIM' comment field to the numbers of the first and last bases of the trimmed sequence (the first base in the sequence is base number zero). Finally set the error probability cutoff value in the `TRIM' comment field to -1.00 to indicate that the sequence is trimmed, and that the trim points may be unrelated to the error probability cutoff value. -trim_out Trim information in the FASTA, SCF, and PHD output files. This is equivalent to specifying `-trim_fasta', `-trim_scf', and `-trim_phd' on the command line. -nonorm Disable phred trace normalization. This option is not recommended unless the base caller fails due to huge noise peaks extending over a large region at the start of the trace, as is characteristic of some dye terminator reactions. -nosplit Disable compressed peak splitting. By default, phred identifies and splits C and G peaks that may be a merged pair of peaks. Phred searches for compression prone loop/stem sequence motifs and attempts to confirm a compression using characteristics of the trace, primarily the size of the candidate peak. -nocmpqv Force phred to use the four parameter quality values. By default, phred uses five parameter quality values for dye primer data (only) in order to reduce the quality values of merged CC and GG peaks. (Phred uses the four parameter quality values for dye terminator chemistry data automatically. If phred cannot determine the chemistry, it uses the four parameter quality values.) -ceilqv Specifies a maximum quality value assigned to bases. Bases with quality value parameters that correspond to quality values greater than are assigned the value . -beg_pred Specifies the trace point at which to begin the peak prediction. This point should be in a region of `good' trace where the peak spacing is even and representative of the peak spacing throughout the trace. In addition the peaks should be large and the noise low in the region, and the value of must not be within 100 points of the trace ends. Output Options -------------- -st fasta Set the output sequence file format to FASTA. (Default.) Trimming options affect the FASTA file; see the Notes below for more information. -st xbap Set the output sequence file format to XBAP. -s Write sequence output files with the names obtained by appending ".seq" to the names of the input files, and store them in the directory where phred is running. -s Write a sequence output file with the name . This option is valid for a single input file only. -sd Write sequence output files with the names obtained by appending ".seq" to the names of the input files, and write them in the directory . -sa Write a sequence output file in FASTA format with the name . The file contains the base calls of all the reads processed in this run of phred. -qt fasta Set the output quality file format to FASTA. Trimming options affect the FASTA file; see the Notes below for more information. -qt xbap Set the output quality file format to XBAP. Trimmed off base quality values are omitted. -qt mix Set the output quality file format to FASTA. Base quality values for all bases are written (including those for trimmed off bases). -q Write quality output files with the names obtained by appending ".qual" to the names of the input files, and store them in the directory where phred is running. This option is valid for FASTA format output files only. -q Write a quality output file with the name . This option is valid for a single input file and a FASTA format output file only. -qd Write quality output files with the names obtained by appending ".qual" to the names of the input files, and store them in the directory . -qa Write a quality output file in FASTA format with the name . The file contains the quality values of all the reads processed in this run of phred. -qr Write a histogram of the number of high quality bases per read. This is meaning- ful when phred processes more than one read. -c Write SCF files with the trace data, the base calls of the current sequences, and the positions of the base calls. The SCF files have the names of the input files (phred will refuse to write the SCF file if you ask it to write the SCF file in the directory in which the input file resides). -c Write an SCF file with the trace data, the base calls of the current sequence, and the positions of the base calls. The SCF file has the name . This option is valid for a single input file only. -cd Write SCF files with the trace data, the base calls of the current sequences, and the positions of the base calls. The SCF files are written in the directory and have the same names as the input files. -cp Store SCF trace data as 1 or 2 byte values. Defaults to 1 when the maximum trace value is less than 256, or to 2 when the maximum trace value is greater than or equal to 256. This is the trace precision. -cs Always scale traces before writing them to an SCF output file. This ensures that the largest trace value has the largest value that can be stored in the SCF file. When the file trace precision is `1', the maximum value is 255, and when the precision is 2, the maximum value is 65535. Without this option, phred does not scale the trace unless (a) the trace was read from an ESD file or (b) the maximum trace value exceeds the value that can be stored in the SCF file at the precision used. Trace scaling ensures the maximum digital resolution for a given storage precision but it will make a uniformly low level trace appear to be a high level. -p Write a PHD file, which is used by the consed editor to display bases. A PHD file contains a set of comments used by consed for maintaining consistency between the chromat file, the .ace file and the PHD file, and it contains base data as triples consisting of the base call, quality, and position. Phred always writes the first version of the PHD file for a read, which has the name .phd.1. When a read is edited using consed, a new version of the phd is written by consed, for example, the second version has the name .phd.2. With the -p option, is the name of the input file. -p Write a PHD file with the name .phd.1. This option is valid for processing a single input file. -pd Write PHD files in directory . The PHD files have the names .phd.1 where is the name of the input file. -d Write a data file that is used for detecting polymorphic bases. The file has the name .poly where is the name of the input file. The first line of the file consists of the sequence name, the smallest amplitude normalization factor, and the amplitude normalization factors for the A, C, G, and T traces. One line for each called base follows the header line. The information on each line consists of the called base, the position of the called base, the area of the called peak, the relative area of the called peak, the uncalled base, the position of the uncalled base, the area of the uncalled base, the relative area of the uncalled base, and the amplitudes of the four traces at the position of the called base. -dd Write polymorphism data files in directory . The files have the names .poly where is the name of the input file. -raw Write in the header of the sequence output file and the quality output file. By default, the name of the input file is written in the headers of these files. This option is valid for a single input file only. -log Make phred append a log entry describing the processing run in the file "phred.log". Miscellaneous ------------- -v Verbose operation. You can control the level of verbosity with , which ranges from 1 to 63. -tags Label common output with tags in order to facilitate output parsing. -h, -help Display a command line option summary. -doc Display phred documentation. -V Display phred version. Examples -------- If you plan to use phred base calls and base quality information as input to the phrap assembly program and to the consed finishing program, simply follow the documentation supplied with consed and then type: phredPhrap (with no arguments) If you intend to use consed, you *MUST* use this perl script. Failure to use this script will result in many consed features not working correctly, including consed's autofinish function, user-defined consensus tags, tagging ALU and other repeats, and tagging vector sequence. Use the phredPhrap perl script. An outline of the important processing steps performed by the script follows. Let us say you want to call bases from the chromat files in subdirectory "chromat_dir", use phrap to assemble the contigs, and run consed to edit/examine the contigs. In this case you must ask phred to create "phd" output files, which are required by consed. It runs phred with the options % phred -id chromat_dir -pd phd_dir which causes phred to read the chromat files in "chromat_dir" and write the "phd" files to "phd_dir". Next it makes FASTA files from the "phd" files by running the phd2fasta program. For example, % phd2fasta -id phd_dir -os seqs_fasta -oq seqs_fasta.screen.qual Subsequently it screens out the vector in the sequences in "seqs_fasta" using cross_match: % cross_match seqs_fasta vector.seq -minmatch 12 -minscore 20 -screen > screen.out which generates the screened sequence file "seqs_fasta.screen", It runs phrap to perform the sequence assembly as follows: % phrap seqs_fasta.screen -new_ace > phrap.out Phrap writes the the assembled contigs to the file "seqs_fasta.screen.contigs", and creates a .ace file that can be used for importing the assembly to xbap, consed, or ace-mbly for editing. As another example, again you want to process the chromat files in subdirectory "chromat_dir", but now you want phred to write the base calls to a FASTA file named "seqs_fasta" and the base quality values to "seqs_fasta.qual". In this case you run phred with the options % phred -id chromat_dir -sa seqs_fasta -qa seqs_fasta.qual We recommend that you not use the trim option. Inaccurate bases called near the ends of the traces will not interfere with proper phrap assembly. Refer to the file "phrap.doc", which is part of the phrap distribution, for information on cross_match and phrap. Return values ------------- Phred returns 0 for successful processing and for file read errors. It returns -1 for processing errors and file write errors. Phred continues processing on file read and write errors but halts on serious processing errors. 6. Phred parameter file Phred reads the `primer ID' information in the chromatogram and it tries to find the same name in the phred parameter file, which is described in the `Building and installing' section above. If it succeeds, the phredpar.dat entry for the `primer ID' identifies the sequencing reaction chemistry (primer or terminator) and the type of dye. If it cannot find the `primer ID' information in the chromatogram, it reports no dye primer ID in chromat yyyy where yyyy is the chromatogram name. If it cannot find the `primer ID' name in phredpar.dat (or it cannot find the phredpar.dat file), it reports unknown chemistry (xxxx) in chromat yyyy add a line of the form "xxxx" to the file zzzz type `phred -doc' for more information where xxxx is the `primer ID' and yyyy is the chromatogram name. Add the indicated line to phredpar.dat. Phred reads the `PHRED_PARAMETER_FILE' environment variable in order to find the phredpar.dat file. If this is not set on your system, phred reports warning: 'PHRED_PARAMETER_FILE' environment variable not set: unable to identify chemistry and dye type `phred -doc' for more information If the `PHRED_PARAMETER_FILE' environment variable is set incorrectly, that is, phred cannot find the phredpar.dat file there or the file is not valid, phred reports readParamFile: warning: unable to open file zzzz warning: processing without phred parameters where zzzz is the value of the PHRED_PARAMETER_FILE environment. It processes the chromatograms but warns that it could not read the parameter file as it processes each chromatogram as explained above. In this case, you must set PHRED_PARAMETER_FILE to a valid name. In these three cases phred processes the chromatogram but it uses the default (ABI) four parameter quality values, does not try to split compression peaks, and reports the chemistry and dye types to phrap as `unknown'. If you use a `primer ID' for your reactions that is not in phredpar.dat, you can add the `primer ID' name to phredpar.dat. You will need to know the `primer ID' name as it is stored in the chromatograms, the chemistry type (primer or terminator), the dye name, and the type of sequencing machine. Use a text editor to add `primer ID' entries to phredpar.dat. You will find additional information about the form of phredpar.dat entries in phredpar.dat. The columns in phredpar.dat have the form column value name ------ ---------- 1 primer identification string 2 chemistry 3 dye 4 sequencing machine type where the column values are separated by spaces or horizontal tabs. The values phred recognizes are value name values ---------- ------ primer id. string primer name enclosed in double quotes chemistry primer, terminator dye rhodamine, d-rhodamine, big-dye, energy-transfer, bodipy sequencing machine type ABI_373_377, MolDyn_MegaBACE, ABI_3700 NOTE: the `MegaBACE Mobility File' entry in the phredpar.dat file specifies `unknown' chemistry, rather than `primer' or `terminator' because some early MegaBACE software wrote `MegaBACE Mobility File' for the `primer ID' string in both primer and terminator chemistry ABD files. You may want to change this value if you process exclusively primer or terminator chemistry MegaBACE data; however, you must remember to change it if you decide to process different chemistry data from the MegaBACE later. 7. Notes Sequence Trimming ----------------- First, a warning: do not trim sequences that phrap will assemble. We cautiously introduce trimming capabilities in phred to allow identification of the high quality region of reads, and to permit trimming off low quality segments of reads that are not destined for a phrap assembly. Phred uses a number of different algorithms to calculate trimming information. The algorithm used and its effect depend on the output file and the trimming-related command line options. The phd output file always contains trimming information in the header. Phred calculates this trimming information using a modified Mott algorithm (it does not trim off vector sequence so the trimming information identifies the entire high quality segment of the read). The trimming information appears in the phd file header in the form TRIM: where is the first high quality base (where the first base in the sequence is number zero) and is the last high quality base. is the error probability cutoff value used to calculate the trim points. The command line option `-trim_cutoff' affects the phd file trimming information by setting the error probability cutoff value used to calculate the base scores. If the sequence has fewer than 20 high quality bases, the values and are set to -1. If the `-trim_phd' or `-trim_out' option is used, and are set to the numbers of the first and last bases in the trimmed sequence (so is always zero), and is set to -1.00 to indicate that the sequence is trimmed and that the error probability cutoff value may be unrelated to the trim points. The sequence, quality value, SCF, and PHD output files can be affected by the trimming-related command line options. (Sequence and quality value files are those created using the -s, -sa, -sd, -q, -qa, and -qd options, SCF files are created using the -c and -cd options, and PHD files are created using the -p and -pd options). When phred runs without trimming-related options set, it does not calculate trimming values for the sequence, quality value, and SCF output files (and it does not `trim' the values stored in them). The `-trim' and `-trim_alt' options select the trimming algorithm used to calculate the trimming information used in the sequence, quality value, and SCF output files. The algorithm used for the `-trim' option is based directly on characteristics of the trace. It predates phred and phred quality values. The algorithm used for the `-trim_alt' option is based on the modified Mott algorithm: it uses the base error probabilities calculated from the phred quality values and the error probability cutoff (the cutoff can be adjusted using the -trim_cutoff option). We believe that the `-trim' option tends to be conservative, `trimming off' more bases, in comparison to the `-trim_alt' option. So we recommend using the `-trim_alt' algorithm. Both the `-trim' and `-trim_alt' options take an argument consisting of a vector sequence. If the argument is "" (null), phred finds the high quality segment of the read. If the argument is not null, and phred finds the beginning of the vector sequence within the first 100 bases of the read, phred sets the left trim point to remove the vector sequence as well as low quality bases. Selecting either `-trim' or `-trim_alt' causes phred to determine trimming information and to modify the sequence, quality value, and SCF files as follows. The FASTA sequence header contains trimming information but the sequence is unaffected. The header has the form >chromat_name 1323 15 548 ABI where the sequence name immediately follows the header delimiter, which is ">", the first integer is the number of bases called by phred, the second integer is the number of bases `trimmed off' the beginning of the sequence, the third integer is the number of bases `remaining following trimming', and the string describes the type of input file. The XBAP-type of sequence header contains trimming information, and the low quality bases are commented out. For quality value file type option `-qt fasta' (default), the FASTA quality value header contains the same trimming information as in the FASTA sequence header and the quality values of the `trimmed off' bases are set to zero. For quality value file type option `-qt xbap', phred writes a XBAP-type of sequence header with trimming information followed by the quality values of the bases remaining after trimming on subsequent lines. For quality value file type option `-qt mix', phred writes a FASTA quality value header with the same trimming information as in the FASTA sequence header followed by the quality values of all bases (without trimming). The SCF file contains trimming information in the header, and the sequence, quality values, and trace locations of the called peaks are unaffected. The left clip is the number of bases to trim off the left end of the sequence and the right clip is the number of bases to trim off the right end. When the `-trim_fasta' or `-trim_out' option is used with the `-trim' or `-trim_alt' (and -s, -sa, -sd, -q, -qa, or -qd) option, phred writes the trimmed sequence to the sequence FASTA file and trimmed quality values to the quality value FASTA file; that is, it writes only the high quality bases and the corresponding quality values. In addition, it appends the string `trimmed' to the FASTA headers and the trimming information in the header indicates that no (additional) bases are to be trimmed off. The option `-trim_fasta' is invalid with the `-qt xbap' and `-qt mix' options. When the `-trim_scf' or `-trim_out' option is used with the `-trim' or `-trim_alt' (and -c or -cd) option, phred writes the trimmed sequence, trimmed quality value, and trimmed called peak locations to the SCF output file. In addition, it appends the string `trimmed' to the comment field and the left and right clip values are set to zero. When the `-trim_phd' or `-trim_out' option is used with the `-trim' or `-trim_alt' (and -p or -pd) option, phred writes the trimmed sequence, trimmed quality value, and trimmed called peak locations to the PHD output file. In addition, when it writes the `TRIM' field in the comment block (at the beginning of the file), it sets the values for the first and last high quality bases to the numbers of the first and last bases of the trimmed sequence (where the first base is number zero), and it sets the error probability cutoff value to -1.00. Setting the cutoff value to -1.00 indicates that the sequence is trimmed, and that the trim points may be unrelated to the error probability cutoff value. The modified Mott trimming algorithm, which is used to calculate the trimming information for the `-trim_alt' option and the phd files, uses base error probabilities calculated from the phred quality values. For each base it subtracts the base error probability from an error probability cutoff value (0.05 by default, and changed using the `-trim_cutoff' option) to form the base score. Then it finds the highest scoring segment of the sequence where the segment score is the sum of the segment base scores (the score can have non-negative values only). The algorithm requires a minimum segment length, which is set to 20 bases. ESD Files --------- Phred reads processed MegaBACE ESD files. It cannot read the raw ESD files. It is important that you identify the dye chemistry correctly when you run the MegaBACE base caller so that phred can assign the right base to each trace. (This is important with ABI data too.) In order to obtain the best phred quality value accuracy with MegaBACE data, phred must use the quality value lookup tables designed for this data. Phred identifies the sequencing machine by reading the `primer ID' string in the chromatogram and matching it with an entry in the phredpar.dat file. The matching entry lists the chemistry, dye, and sequencing machine types. For example, the `primer ID' string of the form `ET Primer' identifies a chromatogram as ET dye primer data generated on a MegaBACE sequencing machine. You can check that phred interprets the `primer ID' string correctly by using the `-v 63' option to have phred write diagnostic information to the screen. LI-COR Data ----------- Band Spread Ratio (BSR) Phred reads SCF files created by the LI-COR gel processing software and has quality value lookup tables calibrated for traces processed with Band Spread Ratio (BSR) of 2.2. The LI-COR software writes a `primer ID' string in the SCF file that indicates the BSR value used in the trace processing, which for BSR=2.2 is `DyePrimer{LI-COR_IR_2.2}'. Accordingly, the phredpar.dat file in this distrino input files specified bution has an entry with this string, which enables phred to recognize LI-COR traces processed with BSR=2.2, and to use the quality value lookup table designed for this LI-COR data. Phred has a quality value lookup table for data processed with BSR=2.2 only so the quality values for LI-COR traces processed with other BSR values will have reduced accuracy. 8. References Brent Ewing, LaDeana Hillier, Michael C. Wendl, and Phil Green. Base-calling of automated sequencer traces using phred. I. Accuracy assessment. 1998. Genome Research 8:175-185. Brent Ewing and Phil Green Base-calling of automated sequencer traces using phred. II. Error probabilities. 1998. Genome Research 8:186-194.