Introduction¶

Pysam is a python module that makes it easy to read and manipulate mapped short read sequence data stored in SAM/BAM files. It is a lightweight wrapper of the htslib C-API.

This page provides a quick introduction in using pysam followed by the API. See Working with BAM/CRAM/SAM-formatted files for more detailed usage instructions.

To use the module to read a file in BAM format, create a AlignmentFile object:

import pysam
samfile = pysam.AlignmentFile("ex1.bam", "rb")

Once a file is opened you can iterate over all of the read mapping to a specified region using fetch(). Each iteration returns a AlignedSegment object which represents a single read along with its fields and optional tags:

for read in samfile.fetch('chr1', 100, 120):
    print(read)

samfile.close()

To give:

EAS56_57:6:190:289:82       0       99      <<<7<<<;<<<<<<<<8;;<7;4<;<;;;;;94<;     69      CTCAAGGTTGTTGCAAGGGGGTCTATGTGAACAAA     0       192     1
EAS56_57:6:190:289:82       0       99      <<<<<<;<<<<<<<<<<;<<;<<<<;8<6;9;;2;     137     AGGGGTGCAGAGCCGAGTCACGGGGTTGCCAGCAC     73      64      1
EAS51_64:3:190:727:308      0       102     <<<<<<<<<<<<<<<<<<<<<<<<<<<::<<<844     99      GGTGCAGAGCCGAGTCACGGGGTTGCCAGCACAGG     99      18      1
...

You can also write to a AlignmentFile:

import pysam
samfile = pysam.AlignmentFile("ex1.bam", "rb")
pairedreads = pysam.AlignmentFile("allpaired.bam", "wb", template=samfile)
for read in samfile.fetch():
    if read.is_paired:
        pairedreads.write(read)

pairedreads.close()
samfile.close()

An alternative way of accessing the data in a SAM file is by iterating over each base of a specified region using the pileup() method. Each iteration returns a PileupColumn which represents all the reads in the SAM file that map to a single base in the reference sequence. The list of reads are represented as PileupRead objects in the PileupColumn.pileups property:

import pysam
samfile = pysam.AlignmentFile("ex1.bam", "rb" )
for pileupcolumn in samfile.pileup("chr1", 100, 120):
    print("\ncoverage at base %s = %s" % (pileupcolumn.pos, pileupcolumn.n))
    for pileupread in pileupcolumn.pileups:
        if not pileupread.is_del and not pileupread.is_refskip:
            # query position is None if is_del or is_refskip is set.
            print('\tbase in read %s = %s' %
                  (pileupread.alignment.query_name,
                   pileupread.alignment.query_sequence[pileupread.query_position]))

samfile.close()

The above code outputs:

coverage at base 99 = 1
    base in read EAS56_57:6:190:289:82 = A

coverage at base 100 = 1
    base in read EAS56_57:6:190:289:82 = G

coverage at base 101 = 1
    base in read EAS56_57:6:190:289:82 = G

coverage at base 102 = 2
    base in read EAS56_57:6:190:289:82 = G
    base in read EAS51_64:3:190:727:308 = G
...

Commands available in samtools are available as simple function calls. For example:

pysam.sort("-o", "output.bam", "ex1.bam")

corresponds to the command line:

samtools sort -o output.bam ex1.bam

Analogous to AlignmentFile, a TabixFile allows fast random access to compressed and tabix indexed tab-separated file formats with genomic data:

import pysam
tabixfile = pysam.TabixFile("example.gtf.gz")

for gtf in tabixfile.fetch("chr1", 1000, 2000):
    print(gtf.contig, gtf.start, gtf.end, gtf.gene_id)

TabixFile implements lazy parsing in order to iterate over large tables efficiently.

More detailed usage instructions is at Working with BAM/CRAM/SAM-formatted files.

Note

Coordinates in pysam are always 0-based (following the python convention). SAM text files use 1-based coordinates.

Note

The above examples can be run in the tests directory of the: installation directory. Type ‘make’ before running them.

API¶

SAM/BAM/CRAM files¶

Objects of type AlignmentFile allow working with BAM/SAM formatted files.

class pysam.AlignmentFile¶

AlignmentFile(filepath_or_object, mode=None, template=None, reference_names=None, reference_lengths=None, text=NULL, header=None, add_sq_text=False, check_header=True, check_sq=True, reference_filename=None, filename=None, index_filename=None, filepath_index=None, require_index=False, duplicate_filehandle=True, ignore_truncation=False, threads=1)

A SAM/BAM/CRAM formatted file.

If filepath_or_object is a string, the file is automatically opened. If filepath_or_object is a python File object, the already opened file will be used.

If the file is opened for reading and an index exists (if file is BAM, a .bai file or if CRAM a .crai file), it will be opened automatically. index_filename may be specified explicitly. If the index is not named in the standard manner, not located in the same directory as the BAM/CRAM file, or is remote. Without an index, random access via fetch() and pileup() is disabled.

For writing, the header of a SAM file/BAM file can be constituted from several sources (see also the samtools format specification):

If template is given, the header is copied from another AlignmentFile (template must be a AlignmentFile).

If header is given, the header is built from a multi-level dictionary.

If text is given, new header text is copied from raw text.

The names (reference_names) and lengths (reference_lengths) are supplied directly as lists.

When reading or writing a CRAM file, the filename of a FASTA-formatted reference can be specified with reference_filename.

By default, if a file is opened in mode ‘r’, it is checked for a valid header (check_header = True) and a definition of chromosome names (check_sq = True).

Parameters:

mode (string) –
mode should be r for reading or w for writing. The default is text mode (SAM). For binary (BAM) I/O you should append b for compressed or u for uncompressed BAM output. Use h to output header information in text (TAM) mode. Use c for CRAM formatted files.

If b is present, it must immediately follow r or w. Valid modes are r, w, wh, rb, wb, wbu, wb0, rc and wc. For instance, to open a BAM formatted file for reading, type:
```
f = pysam.AlignmentFile('ex1.bam','rb')
```
If mode is not specified, the method will try to auto-detect in the order ‘rb’, ‘r’, thus both the following should work:
```
f1 = pysam.AlignmentFile('ex1.bam')
f2 = pysam.AlignmentFile('ex1.sam')
```
template (AlignmentFile) – when writing, copy header from file template.
header (dict or AlignmentHeader) – when writing, build header from a multi-level dictionary. The first level are the four types (‘HD’, ‘SQ’, …). The second level are a list of lines, with each line being a list of tag-value pairs. The header is constructed first from all the defined fields, followed by user tags in alphabetical order. Alternatively, an AlignmentHeader object can be passed directly.
text (string) – when writing, use the string provided as the header
reference_names (list) – see reference_lengths
reference_lengths (list) – when writing or opening a SAM file without header build header from list of chromosome names and lengths. By default, ‘SQ’ and ‘LN’ tags will be added to the header text. This option can be changed by unsetting the flag add_sq_text.
add_sq_text (bool) – do not add ‘SQ’ and ‘LN’ tags to header. This option permits construction SAM formatted files without a header.
add_sam_header (bool) – when outputting SAM the default is to output a header. This is equivalent to opening the file in ‘wh’ mode. If this option is set to False, no header will be output. To read such a file, set check_header=False.
check_header (bool) – obsolete: when reading a SAM file, check if header is present (default=True)
check_sq (bool) – when reading, check if SQ entries are present in header (default=True)
reference_filename (string) – Path to a FASTA-formatted reference file. Valid only for CRAM files. When reading a CRAM file, this overrides both $REF_PATH and the URL specified in the header (UR tag), which are normally used to find the reference.
index_filename (string) – Explicit path to the index file. Only needed if the index is not named in the standard manner, not located in the same directory as the BAM/CRAM file, or is remote. An IOError is raised if the index cannot be found or is invalid.
filepath_index (string) – Alias for index_filename.
require_index (bool) – When reading, require that an index file is present and is valid or raise an IOError. (default=False)
filename (string) – Alternative to filepath_or_object. Filename of the file to be opened.
duplicate_filehandle (bool) – By default, file handles passed either directly or through File-like objects will be duplicated before passing them to htslib. The duplication prevents issues where the same stream will be closed by htslib and through destruction of the high-level python object. Set to False to turn off duplication.
ignore_truncation (bool) – Issue a warning, instead of raising an error if the current file appears to be truncated due to a missing EOF marker. Only applies to bgzipped formats. (Default=False)
format_options (list) – A list of key=value strings, as accepted by –input-fmt-option and –output-fmt-option in samtools.
threads (integer) – Number of threads to use for compressing/decompressing BAM/CRAM files. Setting threads to > 1 cannot be combined with ignore_truncation. (Default=1)

check_index¶

return True if index is present.

Raises:	`AttributeError` – if htsfile is SAM formatted and thus has no index. `ValueError` – if htsfile is closed or index could not be opened.

close¶: closes the pysam.AlignmentFile.

count¶

count the number of reads in region

The region is specified by contig, start and stop. reference and end are also accepted for backward compatibility as synonyms for contig and stop, respectively. Alternatively, a samtools region string can be supplied.

A SAM file does not allow random access and if region or contig are given, an exception is raised.

Parameters:

contig (string) – reference_name of the genomic region (chromosome)
start (int) – start of the genomic region (0-based inclusive)
stop (int) – end of the genomic region (0-based exclusive)
region (string) – a region string in samtools format.
until_eof (bool) – count until the end of the file, possibly including unmapped reads as well.
read_callback (string or function) –
select a call-back to ignore reads when counting. It can be either a string with the following values:

all

skip reads in which any of the following flags are set: BAM_FUNMAP, BAM_FSECONDARY, BAM_FQCFAIL, BAM_FDUP

nofilter

uses every single read

Alternatively, read_callback can be a function check_read(read) that should return True only for those reads that shall be included in the counting.
reference (string) – backward compatible synonym for contig
end (int) – backward compatible synonym for stop

Raises:

ValueError – if the genomic coordinates are out of range or invalid.

count_coverage¶

count the coverage of genomic positions by reads in region.

The region is specified by contig, start and stop. reference and end are also accepted for backward compatibility as synonyms for contig and stop, respectively. Alternatively, a samtools region string can be supplied. The coverage is computed per-base [ACGT].

Parameters:	contig (string) – reference_name of the genomic region (chromosome) start (int) – start of the genomic region (0-based inclusive). If not given, count from the start of the chromosome. stop (int) – end of the genomic region (0-based exclusive). If not given, count to the end of the chromosome. region (string) – a region string. quality_threshold (int) – quality_threshold is the minimum quality score (in phred) a base has to reach to be counted. read_callback (string or function) – select a call-back to ignore reads when counting. It can be either a string with the following values: `all` skip reads in which any of the following flags are set: BAM_FUNMAP, BAM_FSECONDARY, BAM_FQCFAIL, BAM_FDUP `nofilter` uses every single read Alternatively, read_callback can be a function `check_read(read)` that should return True only for those reads that shall be included in the counting. reference (string) – backward compatible synonym for contig end (int) – backward compatible synonym for stop
Raises:	`ValueError` – if the genomic coordinates are out of range or invalid.
Returns:	four array.arrays of the same length in order A C G T
Return type:	tuple

fetch¶

fetch reads aligned in a region.

See parse_region() for more information on how genomic regions can be specified. reference and end are also accepted for backward compatibility as synonyms for contig and stop, respectively.

Without a contig or region all mapped reads in the file will be fetched. The reads will be returned ordered by reference sequence, which will not necessarily be the order within the file. This mode of iteration still requires an index. If there is no index, use until_eof=True.

If only contig is set, all reads aligned to contig will be fetched.

A SAM file does not allow random access. If region or contig are given, an exception is raised.

Parameters:	until_eof (bool) – If until_eof is True, all reads from the current file position will be returned in order as they are within the file. Using this option will also fetch unmapped reads. multiple_iterators (bool) – If multiple_iterators is True, multiple iterators on the same file can be used at the same time. The iterator returned will receive its own copy of a filehandle to the file effectively re-opening the file. Re-opening a file creates some overhead, so beware.
Returns:	An iterator over a collection of reads.
Return type:	IteratorRow
Raises:	`ValueError` – if the genomic coordinates are out of range or invalid or the file does not permit random access to genomic coordinates.

find_introns¶

Return a dictionary {(start, stop) – count} Listing the intronic sites in the reads (identified by ‘N’ in the cigar strings), and their support ( = number of reads ).

read_iterator can be the result of a .fetch(…) call. Or it can be a generator filtering such reads. Example samfile.find_introns((read for read in samfile.fetch(…) if read.is_reverse)

find_introns_slow¶

Return a dictionary {(start, stop) – count} Listing the intronic sites in the reads (identified by ‘N’ in the cigar strings), and their support ( = number of reads ).

read_iterator can be the result of a .fetch(…) call. Or it can be a generator filtering such reads. Example samfile.find_introns((read for read in samfile.fetch(…) if read.is_reverse)

get_index_statistics¶

return statistics about mapped/unmapped reads per chromosome as they are stored in the index, similarly to the statistics printed by the samtools idxstats command.

CRAI indexes do not record these statistics, so for a CRAM file with a .crai index the returned statistics will all be 0.

Returns:	a list of records for each chromosome. Each record has the attributes ‘contig’, ‘mapped’, ‘unmapped’ and ‘total’.
Return type:	list

get_reference_length¶: return reference length corresponding to numerical tid

get_reference_name¶: return reference name corresponding to numerical tid

get_tid¶

return the numerical tid corresponding to reference

returns -1 if reference is not known.

getrname¶: deprecated, use get_reference_name() instead

gettid¶: deprecated, use get_tid() instead

has_index¶: return true if htsfile has an existing (and opened) index.

head¶

return an iterator over the first n alignments.

This iterator is is useful for inspecting the bam-file.

Parameters:	multiple_iterators (bool) – is set to True by default in order to avoid changing the current file position.
Returns:	an iterator over a collection of reads
Return type:	IteratorRowHead

is_valid_tid¶

return True if the numerical tid is valid; False otherwise.

Note that the unmapped tid code (-1) counts as an invalid.

lengths¶: tuple of the lengths of the reference sequences. This is a read-only attribute. The lengths are in the same order as pysam.AlignmentFile.references

mapped¶: int with total number of mapped alignments according to the statistics recorded in the index. This is a read-only attribute. (This will be 0 for a CRAM file indexed by a .crai index, as that index format does not record these statistics.)

mate¶

return the mate of pysam.AlignedSegment read.

Note

Calling this method will change the file position. This might interfere with any iterators that have not re-opened the file.

Note

This method is too slow for high-throughput processing. If a read needs to be processed with its mate, work from a read name sorted file or, better, cache reads.

Returns:	the mate
Return type:	AlignedSegment
Raises:	`ValueError` – if the read is unpaired or the mate is unmapped

nocoordinate¶: int with total number of reads without coordinates according to the statistics recorded in the index, i.e., the statistic printed for “*” by the samtools idxstats command. This is a read-only attribute. (This will be 0 for a CRAM file indexed by a .crai index, as that index format does not record these statistics.)

nreferences¶: int with the number of reference sequences in the file. This is a read-only attribute.

pileup¶

perform a pileup within a region. The region is specified by contig, start and stop (using 0-based indexing). reference and end are also accepted for backward compatibility as synonyms for contig and stop, respectively. Alternatively, a samtools ‘region’ string can be supplied.

Without ‘contig’ or ‘region’ all reads will be used for the pileup. The reads will be returned ordered by contig sequence, which will not necessarily be the order within the file.

Note that SAM formatted files do not allow random access. In these files, if a ‘region’ or ‘contig’ are given an exception is raised.

Note

‘all’ reads which overlap the region are returned. The first base returned will be the first base of the first read ‘not’ necessarily the first base of the region used in the query.

Parameters:	truncate (bool) – By default, the samtools pileup engine outputs all reads overlapping a region. If truncate is True and a region is given, only columns in the exact region specified are returned. max_depth (int) – Maximum read depth permitted. The default limit is ‘8000’. stepper (string) – The stepper controls how the iterator advances. Possible options for the stepper are `all` skip reads in which any of the following flags are set: BAM_FUNMAP, BAM_FSECONDARY, BAM_FQCFAIL, BAM_FDUP `nofilter` uses every single read turning off any filtering. `samtools` same filter and read processing as in samtools pileup. For full compatibility, this requires a ‘fastafile’ to be given. The following options all pertain to filtering of the `samtools` stepper. fastafile (`FastaFile` object.) – This is required for some of the steppers. ignore_overlaps (bool) – If set to True, detect if read pairs overlap and only take the higher quality base. This is the default. flag_filter (int) – ignore reads where any of the bits in the flag are set. The default is BAM_FUNMAP \| BAM_FSECONDARY \| BAM_FQCFAIL \| BAM_FDUP. flag_require (int) – only use reads where certain flags are set. The default is 0. ignore_orphans (bool) – ignore orphans (paired reads that are not in a proper pair). The default is to ignore orphans. min_base_quality (int) – Minimum base quality. Bases below the minimum quality will not be output. The default is 13. adjust_capq_threshold (int) – adjust mapping quality. The default is 0 for no adjustment. The recommended value for adjustment is 50. min_mapping_quality (int) – only use reads above a minimum mapping quality. The default is 0. compute_baq (bool) – re-alignment computing per-Base Alignment Qualities (BAQ). The default is to do re-alignment. Realignment requires a reference sequence. If none is present, no realignment will be performed. redo_baq (bool) – recompute per-Base Alignment Quality on the fly ignoring existing base qualities. The default is False (use existing base qualities).
Returns:	an iterator over genomic positions.
Return type:	IteratorColumn

references¶: tuple with the names of reference sequences. This is a read-only attribute

text¶: deprecated, use references and lengths instead

unmapped¶: int with total number of unmapped reads according to the statistics recorded in the index. This number of reads includes the number of reads without coordinates. This is a read-only attribute. (This will be 0 for a CRAM file indexed by a .crai index, as that index format does not record these statistics.)

write¶

write a single pysam.AlignedSegment to disk.

Raises:	`ValueError` – if the writing failed
Returns:	the number of bytes written. If the file is closed, this will be 0.
Return type:	int

class pysam.AlignmentHeader¶

header information for a AlignmentFile object

Parameters:

header_dict (dict) – build header from a multi-level dictionary. The first level are the four types (‘HD’, ‘SQ’, …). The second level are a list of lines, with each line being a list of tag-value pairs. The header is constructed first from all the defined fields, followed by user tags in alphabetical order. Alternatively, an AlignmentHeader object can be passed directly.
text (string) – use the string provided as the header
reference_names (list) – see reference_lengths
reference_lengths (list) – build header from list of chromosome names and lengths. By default, ‘SQ’ and ‘LN’ tags will be added to the header text. This option can be changed by unsetting the flag add_sq_text.
add_sq_text (bool) – do not add ‘SQ’ and ‘LN’ tags to header. This option permits construction SAM formatted files without a header.

as_dict¶: deprecated, use to_dict() instead

copy¶

classmethod from_dict(cls, header_dict)¶

classmethod from_references(cls, reference_names, reference_lengths, text=None, add_sq_text=True)¶

classmethod from_text(cls, text)¶

get¶

get_reference_length¶

get_reference_name¶

get_tid¶

return the numerical tid corresponding to reference

returns -1 if reference is not known.

is_valid_tid¶

return True if the numerical tid is valid; False otherwise.

Note that the unmapped tid code (-1) counts as an invalid.

items¶

iteritems¶

keys¶

lengths¶: tuple of the lengths of the reference sequences. This is a read-only attribute. The lengths are in the same order as pysam.AlignmentFile.references

nreferences¶

int with the number of reference sequences in the file.

This is a read-only attribute.

references¶: tuple with the names of reference sequences. This is a read-only attribute

to_dict¶

return two-level dictionary with header information from the file.

The first level contains the record (HD, SQ, etc) and the second level contains the fields (VN, LN, etc).

The parser is validating and will raise an AssertionError if if encounters any record or field tags that are not part of the SAM specification. Use the pysam.AlignmentFile.text attribute to get the unparsed header.

The parsing follows the SAM format specification with the exception of the CL field. This option will consume the rest of a header line irrespective of any additional fields. This behaviour has been added to accommodate command line options that contain characters that are not valid field separators.

If no @SQ entries are within the text section of the header, this will be automatically added from the reference names and lengths stored in the binary part of the header.

values¶

An AlignedSegment represents an aligned segment within a SAM/BAM file.

class pysam.AlignedSegment(AlignmentHeader header=None)¶

Class representing an aligned segment.

This class stores a handle to the samtools C-structure representing an aligned read. Member read access is forwarded to the C-structure and converted into python objects. This implementation should be fast, as only the data needed is converted.

For write access, the C-structure is updated in-place. This is not the most efficient way to build BAM entries, as the variable length data is concatenated and thus needs to be resized if a field is updated. Furthermore, the BAM entry might be in an inconsistent state.

One issue to look out for is that the sequence should always be set before the quality scores. Setting the sequence will also erase any quality scores that were set previously.

Parameters:	header – `AlignmentHeader` object to map numerical identifiers to chromosome names. If not given, an empty header is created.

aend¶: deprecated, use reference_end instead.

alen¶: deprecated, use reference_length instead.

aligned_pairs¶: deprecated, use get_aligned_pairs() instead.

bin¶: properties bin

blocks¶: deprecated, use get_blocks() instead.

cigar¶: deprecated, use cigarstring or cigartuples instead.

cigarstring¶

the cigar alignment as a string.

The cigar string is a string of alternating integers and characters denoting the length and the type of an operation.

Note

The order length,operation is specified in the SAM format. It is different from the order of the cigar property.

Returns None if not present.

To unset the cigarstring, assign None or the empty string.

cigartuples¶

the cigar alignment. The alignment is returned as a list of tuples of (operation, length).

If the alignment is not present, None is returned.

The operations are:

M	BAM_CMATCH	0
I	BAM_CINS	1
D	BAM_CDEL	2
N	BAM_CREF_SKIP	3
S	BAM_CSOFT_CLIP	4
H	BAM_CHARD_CLIP	5
P	BAM_CPAD	6
=	BAM_CEQUAL	7
X	BAM_CDIFF	8
B	BAM_CBACK	9

Note

The output is a list of (operation, length) tuples, such as [(0, 30)]. This is different from the SAM specification and the cigarstring property, which uses a (length, operation) order, for example: 30M.

To unset the cigar property, assign an empty list or None.

compare¶: return -1,0,1, if contents in this are binary <,=,> to other

flag¶: properties flag

classmethod from_dict(cls, sam_dict, AlignmentHeader header)¶

parses a dictionary representation of the aligned segment.

Parameters:	sam_dict – dictionary of alignment values, keys corresponding to output from `todict()`.

classmethod fromstring(cls, sam, AlignmentHeader header)¶

parses a string representation of the aligned segment.

The input format should be valid SAM format.

Parameters:	sam – SAM formatted string

get_aligned_pairs¶

a list of aligned read (query) and reference positions.

Each item in the returned list is a tuple consisting of the 0-based offset from the start of the read sequence followed by the 0-based reference position.

For inserts, deletions, skipping either query or reference position may be None.

For padding in the reference, the reference position will always be None.

Parameters:	matches_only (bool) – If True, only matched bases are returned — no None on either side. with_seq (bool) – If True, return a third element in the tuple containing the reference sequence. For CIGAR ‘P’ (padding in the reference) operations, the third tuple element will be None. Substitutions are lower-case. This option requires an MD tag to be present.
Returns:	aligned_pairs
Return type:	list of tuples

get_blocks¶

a list of start and end positions of aligned gapless blocks.

The start and end positions are in genomic coordinates.

Blocks are not normalized, i.e. two blocks might be directly adjacent. This happens if the two blocks are separated by an insertion in the read.

get_cigar_stats¶

summary of operations in cigar string.

The output order in the array is “MIDNSHP=X” followed by a field for the NM tag. If the NM tag is not present, this field will always be 0.

M	BAM_CMATCH	0
I	BAM_CINS	1
D	BAM_CDEL	2
N	BAM_CREF_SKIP	3
S	BAM_CSOFT_CLIP	4
H	BAM_CHARD_CLIP	5
P	BAM_CPAD	6
=	BAM_CEQUAL	7
X	BAM_CDIFF	8
B	BAM_CBACK	9
NM	NM tag	10

If no cigar string is present, empty arrays will be returned.

Returns:	two arrays. The first contains the nucleotide counts within each cigar operation, the second contains the number of blocks for each cigar operation.
Return type:	arrays

get_forward_qualities¶

return the original base qualities of the read sequence, in the same format as the query_qualities property.

Reads mapped to the reverse strand have their base qualities stored reversed in the BAM file. This method returns such reads’ base qualities reversed back to their original orientation.

get_forward_sequence¶

return the original read sequence.

Reads mapped to the reverse strand are stored reverse complemented in the BAM file. This method returns such reads reverse complemented back to their original orientation.

Returns None if the record has no query sequence.

get_overlap¶

return number of aligned bases of read overlapping the interval start and end on the reference sequence.

Return None if cigar alignment is not available.

get_reference_positions¶

a list of reference positions that this read aligns to.

By default, this method returns the (0-based) positions on the reference that are within the read’s alignment, leaving gaps corresponding to deletions and other reference skips.

When full_length is True, the returned list is the same length as the read and additionally includes None values corresponding to insertions or soft-clipping, i.e., to bases of the read that are not aligned to a reference position. (See also get_aligned_pairs() which additionally returns the corresponding positions along the read.)

get_reference_sequence¶

return the reference sequence in the region that is covered by the alignment of the read to the reference.

This method requires the MD tag to be set.

get_tag¶

retrieves data from the optional alignment section given a two-letter tag denoting the field.

The returned value is cast into an appropriate python type.

This method is the fastest way to access the optional alignment section if only few tags need to be retrieved.

Possible value types are “AcCsSiIfZHB” (see BAM format specification) as well as additional value type ‘d’ as implemented in htslib.

Parameters:	tag – data tag. with_value_type – Optional[bool] if set to True, the return value is a tuple of (tag value, type code). (default False)
Returns:	A python object with the value of the tag. The type of the object depends on the data type in the data record.
Raises:	`KeyError` – If tag is not present, a KeyError is raised.

get_tags¶

the fields in the optional alignment section.

Returns a list of all fields in the optional alignment section. Values are converted to appropriate python values. For example: [(NM, 2), (RG, "GJP00TM04")]

If with_value_type is set, the value type as encode in the AlignedSegment record will be returned as well:

[(NM, 2, “i”), (RG, “GJP00TM04”, “Z”)]

This method will convert all values in the optional alignment section. When getting only one or few tags, please see get_tag() for a quicker way to achieve this.

has_tag¶: returns true if the optional alignment section contains a given tag.

infer_query_length¶

infer query length from CIGAR alignment.

This method deduces the query length from the CIGAR alignment but does not include hard-clipped bases.

Returns None if CIGAR alignment is not present.

If always is set to True, infer_read_length is used instead. This is deprecated and only present for backward compatibility.

infer_read_length¶

infer read length from CIGAR alignment.

This method deduces the read length from the CIGAR alignment including hard-clipped bases.

Returns None if CIGAR alignment is not present.

inferred_length¶: deprecated, use infer_query_length() instead.

is_duplicate¶: true if optical or PCR duplicate

is_forward¶: true if read is mapped to forward strand (implemented in terms of is_reverse)

is_mapped¶: true if read itself is mapped (implemented in terms of is_unmapped)

is_paired¶: true if read is paired in sequencing

is_proper_pair¶: true if read is mapped in a proper pair

is_qcfail¶: true if QC failure

is_read1¶: true if this is read1

is_read2¶: true if this is read2

is_reverse¶: true if read is mapped to reverse strand

is_secondary¶: true if not primary alignment

is_supplementary¶: true if this is a supplementary alignment

is_unmapped¶: true if read itself is unmapped

isize¶: deprecated, use template_length instead.

mapping_quality¶: mapping quality

mapq¶: deprecated, use mapping_quality instead.

mate_is_forward¶: true if the mate is mapped to forward strand (implemented in terms of mate_is_reverse)

mate_is_mapped¶: true if the mate is mapped (implemented in terms of mate_is_unmapped)

mate_is_reverse¶: true if the mate is mapped to reverse strand

mate_is_unmapped¶: true if the mate is unmapped

modified_bases¶: Modified bases annotations from Ml/Mm tags. The output is Dict[(canonical base, strand, modification)] -> [ (pos,qual), …] with qual being (256*probability), or -1 if unknown. Strand==0 for forward and 1 for reverse strand modification

modified_bases_forward¶: Modified bases annotations from Ml/Mm tags. The output is Dict[(canonical base, strand, modification)] -> [ (pos,qual), …] with qual being (256*probability), or -1 if unknown. Strand==0 for forward and 1 for reverse strand modification. The positions are with respect to the original sequence from get_forward_sequence()

mpos¶: deprecated, use next_reference_start instead.

mrnm¶: deprecated, use next_reference_id instead.

next_reference_id¶: the reference id of the mate/next read.

next_reference_name¶: reference name of the mate/next read (None if no AlignmentFile is associated)

next_reference_start¶: the position of the mate/next read.

opt¶: deprecated, use get_tag() instead.

overlap¶: deprecated, use get_overlap() instead.

pnext¶: deprecated, use next_reference_start instead.

pos¶: deprecated, use reference_start instead.

positions¶: deprecated, use get_reference_positions() instead.

qend¶: deprecated, use query_alignment_end instead.

qlen¶: deprecated, use query_alignment_length instead.

qname¶: deprecated, use query_name instead.

qqual¶: deprecated, use query_alignment_qualities instead.

qstart¶: deprecated, use query_alignment_start instead.

qual¶: deprecated, use query_qualities instead.

query¶: deprecated, use query_alignment_sequence instead.

query_alignment_end¶

end index of the aligned query portion of the sequence (0-based, exclusive)

This the index just past the last base in query_sequence that is not soft-clipped.

query_alignment_length¶

length of the aligned query sequence.

This is equal to query_alignment_end - query_alignment_start

query_alignment_qualities¶

aligned query sequence quality values (None if not present). These are the quality values that correspond to query_alignment_sequence, that is, they exclude qualities of soft clipped bases. This is equal to query_qualities[query_alignment_start:query_alignment_end].

Quality scores are returned as a python array of unsigned chars. Note that this is not the ASCII-encoded value typically seen in FASTQ or SAM formatted files. Thus, no offset of 33 needs to be subtracted.

This property is read-only.

query_alignment_sequence¶

aligned portion of the read.

This is a substring of query_sequence that excludes flanking bases that were soft clipped (None if not present). It is equal to query_sequence[query_alignment_start:query_alignment_end].

SAM/BAM files may include extra flanking bases that are not part of the alignment. These bases may be the result of the Smith-Waterman or other algorithms, which may not require alignments that begin at the first residue or end at the last. In addition, extra sequencing adapters, multiplex identifiers, and low-quality bases that were not considered for alignment may have been retained.

query_alignment_start¶

start index of the aligned query portion of the sequence (0-based, inclusive).

This the index of the first base in query_sequence that is not soft-clipped.

query_length¶

the length of the query/read.

This value corresponds to the length of the sequence supplied in the BAM/SAM file. The length of a query is 0 if there is no sequence in the BAM/SAM file. In those cases, the read length can be inferred from the CIGAR alignment, see pysam.AlignedSegment.infer_query_length().

The length includes soft-clipped bases and is equal to len(query_sequence).

This property is read-only but is updated when a new query sequence is assigned to this AlignedSegment.

Returns 0 if not available.

query_name¶: the query template name (None if not present)

query_qualities¶

read sequence base qualities, including soft clipped bases (None if not present).

Quality scores are returned as a python array of unsigned chars. Note that this is not the ASCII-encoded value typically seen in FASTQ or SAM formatted files. Thus, no offset of 33 needs to be subtracted.

Note that to set quality scores the sequence has to be set beforehand as this will determine the expected length of the quality score array.

This method raises a ValueError if the length of the quality scores and the sequence are not the same.

query_sequence¶

read sequence bases, including soft clipped bases (None if not present).

Assigning to this attribute will invalidate any quality scores. Thus, to in-place edit the sequence and quality scores, copies of the quality scores need to be taken. Consider trimming for example:

q = read.query_qualities
read.query_sequence = read.query_sequence[5:10]
read.query_qualities = q[5:10]

The sequence is returned as it is stored in the BAM file. (This will be the reverse complement of the original read sequence if the mapper has aligned the read to the reverse strand.)

reference_end¶

aligned reference position of the read on the reference genome.

reference_end points to one past the last aligned residue. Returns None if not available (read is unmapped or no cigar alignment present).

reference_id¶: reference ID

Note

This field contains the index of the reference sequence in the sequence dictionary. To obtain the name of the reference sequence, use get_reference_name()

reference_length¶

aligned length of the read on the reference genome.

This is equal to reference_end - reference_start. Returns None if not available.

reference_name¶: reference name

reference_start¶: 0-based leftmost coordinate

rlen¶: deprecated, use query_length instead.

rname¶: deprecated, use reference_id instead.

rnext¶: deprecated, use next_reference_id instead.

seq¶: deprecated, use query_sequence instead.

setTag¶: deprecated, use set_tag() instead.

set_tag¶

sets a particular field tag to value in the optional alignment section.

value_type describes the type of value that is to entered into the alignment record. It can be set explicitly to one of the valid one-letter type codes. If unset, an appropriate type will be chosen automatically based on the python type of value.

An existing value of the same tag will be overwritten unless replace is set to False. This is usually not recommended as a tag may only appear once in the optional alignment section.

If value is None, the tag will be deleted.

This method accepts valid SAM specification value types, which are:

A: printable char
i: signed int
f: float
Z: printable string
H: Byte array in hex format
B: Integer or numeric array

Additionally, it will accept the integer BAM types (‘cCsSI’)

For htslib compatibility, ‘a’ is synonymous with ‘A’ and the method accepts a ‘d’ type code for a double precision float.

When deducing the type code by the python type of value, the following mapping is applied:

i: python int
f: python float
Z: python str or bytes
B: python array.array, list or tuple

Note that a single character string will be output as ‘Z’ and not ‘A’ as the former is the more general type.

set_tags¶

sets the fields in the optional alignment section with a list of (tag, value) tuples.

The value type of the values is determined from the python type. Optionally, a type may be given explicitly as a third value in the tuple, For example:

x.set_tags([(NM, 2, “i”), (RG, “GJP00TM04”, “Z”)]

This method will not enforce the rule that the same tag may appear only once in the optional alignment section.

tags¶: deprecated, use get_tags() instead.

template_length¶: the observed query template length

tid¶: deprecated, use reference_id instead.

tlen¶: deprecated, use template_length instead.

to_dict¶

returns a json representation of the aligned segment.

Field names are abbreviated versions of the class attributes.

to_string¶

returns a string representation of the aligned segment.

The output format is valid SAM format if a header is associated with the AlignedSegment.

tostring¶

deprecated, use to_string() instead.

Parameters:	htsfile – (deprecated) AlignmentFile object to map numerical identifiers to chromosome names. This parameter is present for backwards compatibility and ignored.

class pysam.PileupColumn¶

A pileup of reads at a particular reference sequence position (column). A pileup column contains all the reads that map to a certain target base.

This class is a proxy for results returned by the samtools pileup engine. If the underlying engine iterator advances, the results of this column will change.

get_mapping_qualities¶

query mapping quality scores at pileup column position.

Returns:	a list of quality scores
Return type:	list

get_num_aligned¶

return number of aligned bases at pileup column position.

This method applies a base quality filter and the number is equal to the size of get_query_sequences(), get_mapping_qualities(), etc.

get_query_names¶

query/read names aligned at pileup column position.

Returns:	a list of query names at pileup column position.
Return type:	list

get_query_positions¶

positions in read at pileup column position.

Returns:	a list of read positions
Return type:	list

get_query_qualities¶

query base quality scores at pileup column position.

Returns:	a list of quality scores
Return type:	list

get_query_sequences¶

query bases/sequences at pileup column position.

Optionally, the bases/sequences can be annotated according to the samtools mpileup format. This is the format description from the samtools mpileup tool:

Information on match, mismatch, indel, strand, mapping
quality and start and end of a read are all encoded at the
read base column. At this column, a dot stands for a match
to the reference base on the forward strand, a comma for a
match on the reverse strand, a '>' or '<' for a reference
skip, `ACGTN' for a mismatch on the forward strand and
`acgtn' for a mismatch on the reverse strand. A pattern
`\+[0-9]+[ACGTNacgtn]+' indicates there is an insertion
between this reference position and the next reference
position. The length of the insertion is given by the
integer in the pattern, followed by the inserted
sequence. Similarly, a pattern `-[0-9]+[ACGTNacgtn]+'
represents a deletion from the reference. The deleted bases
will be presented as `*' in the following lines. Also at
the read base column, a symbol `^' marks the start of a
read. The ASCII of the character following `^' minus 33
gives the mapping quality. A symbol `$' marks the end of a
read segment

To reproduce samtools mpileup format, set all of mark_matches, mark_ends and add_indels to True.

Parameters:	mark_matches (bool) – If True, output bases matching the reference as “.” or “,” for forward and reverse strand, respectively. This mark requires the reference sequence. If no reference is present, this option is ignored. mark_ends (bool) – If True, add markers “^” and “$” for read start and end, respectively. add_indels (bool) – If True, add bases for bases inserted into or skipped from the reference. The latter requires a reference sequence file to have been given, e.g. via pileup(fastafile = …). If no reference sequence is available, skipped bases are represented as ‘N’s.
Returns:	a list of bases/sequences per read at pileup column position.
Return type:	list

n¶: deprecated, use nsegments instead.

nsegments¶

number of reads mapping to this column.

Note that this number ignores the base quality filter.

pileups¶: list of reads (pysam.PileupRead) aligned to this column

pos¶: deprecated, use reference_pos instead.

reference_id¶: the reference sequence number as defined in the header

reference_name¶: reference name (None if no AlignmentFile is associated)

reference_pos¶: the position in the reference sequence (0-based).

set_min_base_quality¶: set the minimum base quality for this pileup column.

tid¶: deprecated, use reference_id instead.

class pysam.PileupRead¶

Representation of a read aligned to a particular position in the reference sequence.

alignment¶: a pysam.AlignedSegment object of the aligned read

indel¶

indel length for the position following the current pileup site.

This quantity peeks ahead to the next cigar operation in this alignment. If the next operation is an insertion, indel will be positive. If the next operation is a deletion, it will be negation. 0 if the next operation is not an indel.

is_del¶: 1 iff the base on the padded read is a deletion

is_head¶: 1 iff the base on the padded read is the left-most base.

is_refskip¶: 1 iff the base on the padded read is part of CIGAR N op.

is_tail¶: 1 iff the base on the padded read is the right-most base.

level¶: the level of the read in the “viewer” mode. Note that this value is currently not computed.

query_position¶: position of the read base at the pileup site, 0-based. None if is_del or is_refskip is set.

query_position_or_next¶

position of the read base at the pileup site, 0-based.

If the current position is a deletion, returns the next aligned base.

class pysam.IndexedReads(AlignmentFile samfile, int multiple_iterators=True)¶

Index a Sam/BAM-file by query name while keeping the original sort order intact.

The index is kept in memory and can be substantial.

By default, the file is re-opened to avoid conflicts if multiple operators work on the same file. Set multiple_iterators = False to not re-open samfile.

Parameters:	samfile (AlignmentFile) – File to be indexed. multiple_iterators (bool) – Flag indicating whether the file should be reopened. Reopening prevents existing iterators being affected by the indexing.

build¶: build the index.

find¶

find query_name in index.

Returns:	Returns an iterator over all reads with query_name.
Return type:	IteratorRowSelection
Raises:	`KeyError` – if the query_name is not in the index.

Tabix files¶

TabixFile opens tabular files that have been indexed with tabix.

class pysam.TabixFile¶

Random access to bgzf formatted files that have been indexed by tabix.

The file is automatically opened. The index file of file <filename> is expected to be called <filename>.tbi by default (see parameter index).

Parameters:

filename (string) – Filename of bgzf file to be opened.
index (string) – The filename of the index. If not set, the default is to assume that the index is called filename.tbi
mode (char) – The file opening mode. Currently, only r is permitted.
parser (pysam.Parser) – sets the default parser for this tabix file. If parser is None, the results are returned as an unparsed string. Otherwise, parser is assumed to be a functor that will return parsed data (see for example asTuple and asGTF).
encoding (string) – The encoding passed to the parser
threads (integer) – Number of threads to use for decompressing Tabix files. (Default=1)

Raises:

ValueError – if index file is missing.
IOError – if file could not be opened

close¶: closes the pysam.TabixFile.

contigs¶: list of chromosome names

fetch¶

fetch one or more rows in a region using 0-based indexing. The region is specified by reference, start and end. Alternatively, a samtools region string can be supplied.

Without reference or region all entries will be fetched.

If only reference is set, all reads matching on reference will be fetched.

If parser is None, the default parser will be used for parsing.

Set multiple_iterators to true if you will be using multiple iterators on the same file at the same time. The iterator returned will receive its own copy of a filehandle to the file effectively re-opening the file. Re-opening a file creates some overhead, so beware.

header¶

the file header.

The file header consists of the lines at the beginning of a file that are prefixed by the comment character #.

Note

The header is returned as an iterator presenting lines without the newline character.

To iterate over tabix files, use tabix_iterator():

pysam.tabix_iterator(infile, parser)¶

return an iterator over all entries in a file.

Results are returned parsed as specified by the parser. If parser is None, the results are returned as an unparsed string. Otherwise, parser is assumed to be a functor that will return parsed data (see for example asTuple and asGTF).

pysam.tabix_compress(filename_in, filename_out, force=False)¶

compress filename_in writing the output to filename_out.

Raise an IOError if filename_out already exists, unless force is set.

pysam.tabix_index(filename, force=False, seq_col=None, start_col=None, end_col=None, preset=None, meta_char=u'#', int line_skip=0, zerobased=False, int min_shift=-1, index=None, keep_original=False, csi=False)¶

index tab-separated filename using tabix.

An existing index will not be overwritten unless force is set.

The index will be built from coordinates in columns seq_col, start_col and end_col.

The contents of filename have to be sorted by contig and position - the method does not check if the file is sorted.

Column indices are 0-based. Note that this is different from the tabix command line utility where column indices start at 1.

Coordinates in the file are assumed to be 1-based unless zerobased is set.

If preset is provided, the column coordinates are taken from a preset. Valid values for preset are “gff”, “bed”, “sam”, “vcf”, psltbl”, “pileup”.

Lines beginning with meta_char and the first line_skip lines will be skipped.

If filename is not detected as a gzip file it will be automatically compressed. The original file will be removed and only the compressed file will be retained.

By default or when min_shift is 0, creates a TBI index. If min_shift is greater than zero and/or csi is True, creates a CSI index with a minimal interval size of 1<<min_shift (1<<14 if only csi is set).

index controls the filename which should be used for creating the index. If not set, the default is to append .tbi to filename.

When automatically compressing files, if keep_original is set the uncompressed file will not be deleted.

returns the filename of the compressed data

class pysam.asTuple¶

converts a tabix row into a python tuple.

A field in a row is accessed by numeric index.

class pysam.asVCF¶

converts a tabix row into a VCF record with the following fields:

Column	Field	Contents
1	contig	chromosome
2	pos	chromosomal position, zero-based
3	id	id
4	ref	reference allele
5	alt	alternate alleles
6	qual	quality
7	filter	filter
8	info	info
9	format	format specifier.

Access to genotypes is via index:

contig = vcf.contig
first_sample_genotype = vcf[0]
second_sample_genotype = vcf[1]

class pysam.asBed¶

converts a tabix row into a bed record with the following fields:

Column	Field	Contents
1	contig	contig
2	start	genomic start coordinate (zero-based)
3	end	genomic end coordinate plus one (zero-based)
4	name	name of feature.
5	score	score of feature
6	strand	strand of feature
7	thickStart	thickStart
8	thickEnd	thickEnd
9	itemRGB	itemRGB
10	blockCount	number of bocks
11	blockSizes	‘,’ separated string of block sizes
12	blockStarts	‘,’ separated string of block genomic start positions

Only the first three fields are required. Additional fields are optional, but if one is defined, all the preceding need to be defined as well.

class pysam.asGTF¶

converts a tabix row into a GTF record with the following fields:

Column	Name	Content
1	contig	the chromosome name
2	feature	The feature type
3	source	The feature source
4	start	genomic start coordinate (0-based)
5	end	genomic end coordinate (0-based)
6	score	feature score
7	strand	strand
8	frame	frame
9	attributes	the attribute field

GTF formatted entries also define the following fields that are derived from the attributes field:

Name	Content
gene_id	the gene identifier
transcript_id	the transcript identifier

FASTA files¶

class pysam.FastaFile¶

Random access to fasta formatted files that have been indexed by faidx.

The file is automatically opened. The index file of file <filename> is expected to be called <filename>.fai.

Parameters:	filename (string) – Filename of fasta file to be opened. filepath_index (string) – Optional, filename of the index. By default this is the filename + “.fai”. filepath_index_compressed (string) – Optional, filename of the index if fasta file is. By default this is the filename + “.gzi”.
Raises:	`ValueError` – if index file is missing `IOError` – if file could not be opened

close¶: close the file.

closed¶: bool indicating the current state of the file object. This is a read-only attribute; the close() method changes the value.

fetch¶

fetch sequences in a region.

A region can either be specified by reference, start and end. start and end denote 0-based, half-open intervals.

Alternatively, a samtools region string can be supplied.

If any of the coordinates are missing they will be replaced by the minimum (start) or maximum (end) coordinate.

Note that region strings are 1-based, while start and end denote an interval in python coordinates. The region is specified by reference, start and end.

Returns:	string
Return type:	a string with the sequence specified by the region.
Raises:	`IndexError` – if the coordinates are out of range `ValueError` – if the region is invalid

filename¶: filename associated with this object. This is a read-only attribute.

get_reference_length¶: return the length of reference.

is_open¶: return true if samfile has been opened.

lengths¶: tuple with the lengths of reference sequences.

nreferences¶: int with the number of reference sequences in the file. This is a read-only attribute.

references¶: tuple with the names of reference sequences.

FASTQ files¶

class pysam.FastxFile¶

Stream access to fasta or fastq formatted files.

The file is automatically opened.

Entries in the file can be both fastq or fasta formatted or even a mixture of the two.

This file object permits iterating over all entries in the file. Random access is not implemented. The iteration returns objects of type FastqProxy

Parameters:	filename (string) – Filename of fasta/fastq file to be opened. persist (bool) – If True (default) make a copy of the entry in the file during iteration. If set to False, no copy will be made. This will permit much faster iteration, but an entry will not persist when the iteration continues and an entry is read-only.

Notes

Prior to version 0.8.2, this class was called FastqFile.

Raises:	`IOError` – if file could not be opened

Examples

>>> with pysam.FastxFile(filename) as fh:
...    for entry in fh:
...        print(entry.name)
...        print(entry.sequence)
...        print(entry.comment)
...        print(entry.quality)
>>> with pysam.FastxFile(filename) as fin, open(out_filename, mode='w') as fout:
...    for entry in fin:
...        fout.write(str(entry) + '\n')

close¶: close the file.

closed¶: bool indicating the current state of the file object. This is a read-only attribute; the close() method changes the value.

filename¶: string with the filename associated with this object.

is_open¶: return true if samfile has been opened.

class pysam.FastqProxy¶

A single entry in a fastq file.

get_quality_array¶: return quality values as integer array after subtracting offset.

name¶: The name of each entry in the fastq file.

quality¶: The quality score of each entry in the fastq file, represented as a string.

sequence¶: The sequence of each entry in the fastq file.

VCF/BCF files¶

class pysam.VariantFile(*args, **kwargs)¶

(filename, mode=None, index_filename=None, header=None, drop_samples=False, duplicate_filehandle=True, ignore_truncation=False, threads=1)

A VCF/BCF formatted file. The file is automatically opened.

If an index for a variant file exists (.csi or .tbi), it will be opened automatically. Without an index random access to records via fetch() is disabled.

For writing, a VariantHeader object must be provided, typically obtained from another VCF file/BCF file.

Parameters:

mode (string) –
mode should be r for reading or w for writing. The default is text mode (VCF). For binary (BCF) I/O you should append b for compressed or u for uncompressed BCF output.

If b is present, it must immediately follow r or w. Valid modes are r, w, wh, rb, wb, wbu and wb0. For instance, to open a BCF formatted file for reading, type:
```
f = pysam.VariantFile('ex1.bcf','r')
```
If mode is not specified, we will try to auto-detect the file type. All of the following should work:
```
f1 = pysam.VariantFile('ex1.bcf')
f2 = pysam.VariantFile('ex1.vcf')
f3 = pysam.VariantFile('ex1.vcf.gz')
```
index_filename (string) – Explicit path to an index file.
header (VariantHeader) – VariantHeader object required for writing.
drop_samples (bool) – Ignore sample information when reading.
duplicate_filehandle (bool) – By default, file handles passed either directly or through File-like objects will be duplicated before passing them to htslib. The duplication prevents issues where the same stream will be closed by htslib and through destruction of the high-level python object. Set to False to turn off duplication.
ignore_truncation (bool) – Issue a warning, instead of raising an error if the current file appears to be truncated due to a missing EOF marker. Only applies to bgzipped formats. (Default=False)
threads (integer) – Number of threads to use for compressing/decompressing VCF/BCF files. Setting threads to > 1 cannot be combined with ignore_truncation. (Default=1)

close¶: closes the pysam.VariantFile.

copy¶

fetch¶

fetch records in a region, specified either by contig, start, and end (which are 0-based, half-open); or alternatively by a samtools region string (which is 1-based inclusive).

Without contig or region all mapped records will be fetched. The records will be returned ordered by contig, which will not necessarily be the order within the file.

Set reopen to true if you will be using multiple iterators on the same file at the same time. The iterator returned will receive its own copy of a filehandle to the file effectively re-opening the file. Re-opening a file incurrs some overhead, so use with care.

If only contig is set, all records on contig will be fetched. If both region and contig are given, an exception is raised.

Note that a bgzipped VCF.gz file without a tabix/CSI index (.tbi/.csi) or a BCF file without a CSI index can only be read sequentially.

get_reference_name¶: return reference name corresponding to numerical tid

get_tid¶

return the numerical tid corresponding to reference

returns -1 if reference is not known.

is_valid_tid¶

return True if the numerical tid is valid; False otherwise.

returns -1 if reference is not known.

new_record¶

Create a new empty VariantRecord.

See VariantHeader.new_record()

open¶

open a vcf/bcf file.

If open is called on an existing VariantFile, the current file will be closed and a new file will be opened.

reset¶: reset file position to beginning of file just after the header.

subset_samples¶: Read only a subset of samples to reduce processing time and memory. Must be called prior to retrieving records.

write¶

write a single pysam.VariantRecord to disk.

returns the number of bytes written.

class pysam.VariantHeader¶

header information for a VariantFile object

add_line¶: Add a metadata line to this header

add_meta¶: Add metadata to this header

add_record¶: Add an existing VariantHeaderRecord to this header

add_sample¶: Add a new sample to this header

add_samples¶: Add several new samples to this header. This function takes multiple arguments, each of which may be either a sample name or an iterable returning sample names (e.g., a list of sample names).

alts¶

alt metadata (dict ID->record).

The data returned just a snapshot of alt records, is created every time the property is requested, and modifications will not be reflected in the header metadata and vice versa.

i.e. it is just a dict that reflects the state of alt records at the time it is created.

contigs¶: contig information (VariantHeaderContigs)

copy¶

filters¶: filter metadata (VariantHeaderMetadata)

formats¶: format metadata (VariantHeaderMetadata)

info¶: info metadata (VariantHeaderMetadata)

merge¶

new_record¶

Create a new empty VariantRecord.

Arguments are currently experimental. Use with caution and expect changes in upcoming releases.

records¶: header records (VariantHeaderRecords)

samples¶

version¶: VCF version

class pysam.VariantRecord(*args, **kwargs)¶

Variant record

alleles¶: tuple of reference allele followed by alt alleles

alts¶: tuple of alt alleles

chrom¶: chromosome/contig name

contig¶: chromosome/contig name

copy¶: return a copy of this VariantRecord object

filter¶: filter information (see VariantRecordFilter)

format¶: sample format metadata (see VariantRecordFormat)

id¶: record identifier or None if not available

info¶: info data (see VariantRecordInfo)

pos¶: record start position on chrom/contig (1-based inclusive)

qual¶: phred scaled quality score or None if not available

ref¶: reference allele

rid¶: internal reference id number

rlen¶: record length on chrom/contig (aka rec.stop - rec.start)

samples¶: sample data (see VariantRecordSamples)

start¶: record start position on chrom/contig (0-based inclusive)

stop¶: record stop position on chrom/contig (0-based exclusive)

translate¶

class pysam.VariantHeaderRecord(*args, **kwargs)¶

header record from a VariantHeader object

attrs¶: sequence of additional header attributes

get¶: D.get(k[,d]) -> D[k] if k in D, else d. d defaults to None.

items¶: D.items() -> list of D’s (key, value) pairs, as 2-tuples

iteritems¶: D.iteritems() -> an iterator over the (key, value) items of D

iterkeys¶: D.iterkeys() -> an iterator over the keys of D

itervalues¶: D.itervalues() -> an iterator over the values of D

key¶: header key (the part before ‘=’, in FILTER/INFO/FORMAT/contig/fileformat etc.)

keys¶: D.keys() -> list of D’s keys

pop¶

remove¶

type¶: header type – FILTER, INFO, FORMAT, CONTIG, STRUCTURED, or GENERIC

update¶

D.update([E, ]**F) -> None.

Update D from dict/iterable E and F.

value¶: header value. Set only for generic lines, None for FILTER/INFO, etc.

values¶: D.values() -> list of D’s values

HTSFile¶

HTSFile is the base class for pysam.AlignmentFile and pysam.VariantFile.

class pysam.HTSFile¶

Base class for HTS file types

add_hts_options¶: Given a list of key=value format option strings, add them to an open htsFile

category¶: General file format category. One of UNKNOWN, ALIGNMENTS, VARIANTS, INDEX, REGIONS

check_truncation¶: Check if file is truncated.

close¶

closed¶: return True if HTSFile is closed.

compression¶

File compression.

One of NONE, GZIP, BGZF, CUSTOM.

description¶: Vaguely human readable description of the file format

format¶

File format.

One of UNKNOWN, BINARY_FORMAT, TEXT_FORMAT, SAM, BAM, BAI, CRAM, CRAI, VCF, BCF, CSI, GZI, TBI, BED.

get_reference_name¶: return contig name corresponding to numerical tid

get_tid¶

return the numerical tid corresponding to contig

returns -1 if contig is not known.

is_bam¶: return True if HTSFile is reading or writing a BAM alignment file

is_bcf¶: return True if HTSFile is reading or writing a BCF variant file

is_closed¶: return True if HTSFile is closed.

is_cram¶: return True if HTSFile is reading or writing a BAM alignment file

is_open¶: return True if HTSFile is open and in a valid state.

is_read¶: return True if HTSFile is open for reading

is_sam¶: return True if HTSFile is reading or writing a SAM alignment file

is_valid_reference_name¶: return True if the contig name contig is valid; False otherwise.

is_valid_tid¶

return True if the numerical tid is valid; False otherwise.

returns -1 if contig is not known.

is_vcf¶: return True if HTSFile is reading or writing a VCF variant file

is_write¶: return True if HTSFile is open for writing

parse_region¶

parse alternative ways to specify a genomic region. A region can either be specified by contig, start and stop. start and stop denote 0-based, half-open intervals. reference and end are also accepted for backward compatibility as synonyms for contig and stop, respectively.

Alternatively, a samtools region string can be supplied.

If any of the coordinates are missing they will be replaced by the minimum (start) or maximum (stop) coordinate.

Note that region strings are 1-based inclusive, while start and stop denote an interval in 0-based, half-open coordinates (like BED files and Python slices).

If contig or region or are *, unmapped reads at the end of a BAM file will be returned. Setting either to . will iterate from the beginning of the file.

Returns:	a tuple of flag, tid, start and stop. The flag indicates whether no coordinates were supplied and the genomic region is the complete genomic space.
Return type:	tuple
Raises:	`ValueError` – for invalid or out of bounds regions.

reset¶

reset file position to beginning of file just after the header.

Returns:	The file position after moving the file pointer.
Return type:	pointer

seek¶: move file pointer to position offset, see pysam.HTSFile.tell().

tell¶: return current file position, see pysam.HTSFile.seek().

version¶: Tuple of file format version numbers (major, minor)