Pathogen detection using next generation sequencing

A method for identifying one or more pathogens in a sample of biological material, the method comprising, at a computer system:

• receiving a population of 1 million or more sequence reads as fasta, fastq, or sam files, wherein the sequence reads are obtained from a sequencing of DNA molecules or reverse transcribed RNA molecules (cDNA) from the sample of biological material, the sample including DNA or cDNA molecules from a plurality of organisms;
• using an initial global alignment technique to align the population of sequence reads to a plurality of classified reference genomes in an electronic database, thereby obtaining alignment results that include, for each of at least a portion of the population of sequence reads, at least one matching reference genome to which the population of sequence read aligns, thereby obtaining a set of matching reference genomes, and wherein the alignment results may further include classification information for each of the matching reference genomes, the classification information including a plurality of taxonomy identifiers;
• filtering the alignment results for each matching reference genome of the set of matching reference genomes, by:
• identifying an optimally-aligning sequence read that aligns to the matching reference genome with an optimal alignment score that exceeds or is equal to alignments scores of other sequence reads that align to the matching reference genome, thereby obtaining optimally-aligning sequence reads;
• for each of the optimally-aligning sequence reads:
• applying a second local alignment technique for the optimally-aligning sequence read to the classified reference genomes to obtain a plurality of new alignment scores, wherein the second local alignment technique is at least 1,000 times slower than the initial global alignment technique;
• determining whether any of the new alignment scores exceeds the optimal alignment score for the optimally-aligning sequence read;
• removing a matching reference genome from the set of matching reference genomes and the corresponding sequence reads associated with the matching reference genome, if the new alignment score exceeds the optimal alignment score for the optimally-aligning sequence read;
• identifying a set of the sequence reads that are aligned to two or more of the classified reference genomes with at a least a minimum alignment threshold;
• for each sequence read of the set of the sequence reads:
• identifying two or more matching reference genomes of the classified reference genomes to which the sequence read aligns with at least the minimum alignment threshold;
• assigning a taxonomy identifier to each of the two or more matching reference genomes of the classified reference genomes, thereby obtaining assigned taxonomy identifiers, each taxonomy identifier including at least two levels of classification, the at least two levels having a hierarchy such that there is a lower level and at least one higher level;
• comparing the assigned taxonomy identifiers of the two or more classified reference genomes at each of the at least two levels of classification;
• removing each level of the at least two levels from the assigned taxonomy identifiers that do not match between the two or more classified reference genomes;
• and
• assigning to the sequence read of the set of the sequence reads a lowest level of the at least two levels of the assigned taxonomy identifiers that is shared between the two or more classified reference genomes;
• and
• providing an identification of one or more taxonomy identifiers based on numbers of corresponding sequence reads;
• using sequence reads having an assigned taxonomy identifier to identify the one or more taxonomy identifiers corresponding to the one or more candidate pathogens for each of a set of taxonomy identifiers, by:
• determining a total score based on:
• at least a coverage of a reference genome and a length of the reference genome corresponding to the taxonomy identifier, and
• assigning an identity score corresponding to a fractional sum over genomic positions in the reference genome, the fractional sum of a fraction of sequence reads having an exact match at a genomic position relative to the total number of sequence reads that match or do not match at the genomic position, thereby obtaining total scores for the set of taxonomy identifiers;
• ranking the total scores;
• and
• identifying the one or more taxonomy identifiers exceeding a score threshold.