src/formatdb.c File Reference

#include <blast.h>

Include dependency graph for formatdb.c:

Go to the source code of this file.

Functions
uint4	determineDbAlphabetType (char *filename)
int4	main (int argc, char *argv[])

---

Function Documentation

uint4 determineDbAlphabetType

(

char *

filename

)

Definition at line 128 of file formatdb.c.

References encoding_determineAlphabetType(), encoding_nucleotide, encoding_protein, int4, readFasta_close(), readFasta_open(), readFasta_readSequence(), readFasta_sequenceBuffer, readFasta_sequenceLength, and uint4.

Referenced by main().

{
        int4 sequenceCount = 0;
    char* sequence;
    uint4 sequenceLength;

    // Open FASTA file for reading
        readFasta_open(filename);

        // Move through the FASTA file reading descriptions and sequences
        while (readFasta_readSequence() && sequenceCount < 10)
        {
                // Get sequence just read
                sequence = readFasta_sequenceBuffer;
                sequenceLength = readFasta_sequenceLength;

        // Determine the alphabet of the current sequence
        if (encoding_determineAlphabetType(sequence, sequenceLength) == encoding_protein)
                {
                // If contains protein letters, return protein type
            readFasta_close();
                        return encoding_protein;
        }

        sequenceCount++;
        }

        // Close fasta reader and return nucleotide type
        readFasta_close();
    return encoding_nucleotide;
}

Here is the call graph for this function:

Here is the caller graph for this function:

int4 main	(	int	argc,
		char *	argv[]
	)

Definition at line 12 of file formatdb.c.

References determineDbAlphabetType(), encoding_encodeSequence(), encoding_initialize(), encoding_nucleotide, encoding_protein, encoding_replaceWildcards(), global_realloc(), int4, memSingleBlock_getCurrent(), memSingleBlock_initialize(), memSingleBlock_resetCurrent(), memSingleBlock::numEntries, readFasta_close(), readFasta_descriptionBuffer, readFasta_descriptionLength, readFasta_open(), readFasta_readSequence(), readFasta_sequenceBuffer, readFasta_sequenceLength, uint4, vbyte_putVbyte, writedb_addSequence(), writedb_close(), writedb_initialize(), writedb_maximumSequenceLength, writedb_minimumSequenceLength, writedb_numberOfLetters, writedb_sequenceCount, and writedb_volume.

{
        char *sequence, *filename;
    uint4 sequenceLength;
    int4 totalWilds = 0, alphabetType;
    struct memSingleBlock* wildcardEdits;
    struct wildcardEdit* wildcardEdit;
    char *wildcardData = NULL, *startWildcardData = NULL;

        // User must provide FASTA format file at command line
        if (argc < 2)
        {
                fprintf(stderr, "Useage: formatdb <FASTA file>\n");
                exit(-1);
        }
        filename = argv[1];

    // Initialize array to store wildcard edits
    wildcardEdits = memSingleBlock_initialize(sizeof(struct wildcardEdit), 10);

    // Determine if database is protein or nucleotide
        alphabetType = determineDbAlphabetType(filename);

    if (alphabetType == encoding_protein)
    {
        printf("PROTEIN database detected.\n");
    }
    else if (alphabetType == encoding_nucleotide)
    {
        printf("NUCLEOTIDE database detected.\n");
    }

        // Initialize codes array
        encoding_initialize(alphabetType);

        // Initialize writing to formatted database
    writedb_initialize(filename, alphabetType);

    // Open FASTA file for reading
        readFasta_open(filename);

        printf("Formatting database...");
        fflush(stdout);

        // Move through the FASTA file reading descriptions and sequences
        while (readFasta_readSequence())
        {
                // Get sequence just read
                sequence = readFasta_sequenceBuffer;
                sequenceLength = readFasta_sequenceLength;

        // Encode the sequence
                encoding_encodeSequence(sequence, sequenceLength, alphabetType);

        // Convert nucleotide sequences to byte-packed format
        if (alphabetType == encoding_nucleotide)
        {
                        // Replace any wilds with a random character
            totalWilds += encoding_replaceWildcards(wildcardEdits, sequence, sequenceLength);

                        // Declare memory to hold wildcard data
            startWildcardData = global_realloc(startWildcardData,
                                sizeof(char) * wildcardEdits->numEntries * 5);
            wildcardData = startWildcardData;

            // For each wildcard edit, encode details using chars and vbytes
            memSingleBlock_resetCurrent(wildcardEdits);
            while ((wildcardEdit = memSingleBlock_getCurrent(wildcardEdits)) != NULL)
            {
                // Record wild character
                *wildcardData = wildcardEdit->code;
                wildcardData++;

                // Convert the position to a vbyte
                vbyte_putVbyte(wildcardData, wildcardEdit->position);
            }
                }
        else
        {
                startWildcardData = wildcardData = NULL;
                }

//        printf("[%s](%d)", readFasta_descriptionBuffer, readFasta_descriptionLength); fflush(stdout);

        // Add sequence to the formatted collection
        writedb_addSequence(sequence, sequenceLength, readFasta_descriptionBuffer,
                            readFasta_descriptionLength, startWildcardData,
                            wildcardData - startWildcardData, NULL, 0);

                // Print status dots
                if (writedb_sequenceCount % 10000 == 0)
                {
                        printf(".");
                        fflush(stdout);
                }
        }

        // Close fasta reader
        readFasta_close();

    // Finalize writing to the formatted collection
    writedb_close();

        printf("done.\n");
        printf("%d sequences processed.\n", writedb_sequenceCount);
        printf("%llu letters processed.\n", writedb_numberOfLetters);
    printf("%d wildcards encoded.\n", totalWilds);
        printf("%d volume(s) created.\n", writedb_volume + 1);
        printf("Longest/shortest sequence was %d/%d letters\n",
           writedb_maximumSequenceLength, writedb_minimumSequenceLength);
        fflush(stdout);

        return 0;
}

Here is the call graph for this function:

---