Upgrading: what changed in your results¶

pyMzLib bundles one mzLib build, and every pyMzLib release moves it forward. Most of what a new mzLib brings is new capability. This page is the rest: changes that give you a different answer from the same file and the same call. Read the section for the release you are moving to before you compare numbers across the upgrade.

Each entry names the mzLib pull request it comes from, so you can read the change itself. The changelog has everything else. Which mzLib a bridge was built from is in pymzlib.bridge_version()["mzlib"].

0.4.0 (still built from mzLib 1.0.593)¶

mzLib did not move, so no mzLib change gives you a different answer. These three are pyMzLib's own.

Calls that used to succeed are now refused¶

out= must name a .tsv file on every readers.read_*() function, single-file and _many (any case: .TSV is fine). out="table.csv", or a path with no extension, is now a UsageError before anything is read. The file was always tab-separated, so the fix is to rename it.

Same file, different value¶

read_records() on an SDRF returns no columns. header and cells used to come back as ;-joined strings, which could not be split back because SDRF cells contain ; themselves. Both are now listed in excluded_fields with the reason "use sdrf read"; record_count is unchanged. Read SDRFs with sdrf.read().

Same failure, different error¶

A read whose answer is too large to return (past about 1.07 billion characters of JSON) is a UsageError naming the record count and telling you to use limit/offset or out=. It used to fail with an unexplained OutOfMemoryException.

0.3.0 (built from mzLib 1.0.593)¶

Files that used to be refused now read¶

MetaMorpheus protein-group tables written without quantification (mzLib#1365). AllProteinGroups.tsv and each file's <file>_ProteinGroups.tsv raised Tsv file type not supported. They now read as MetaMorpheusQuantifiedProteinGroups, with spectral counts and no intensities: intensity is in absent_fields, meaning no basis, not zero.
RNA quantification tables (mzLib#1388). AllQuantifiedTranscriptGroups.tsv and AllQuantifiedOligos.tsv are new file types, so formats() lists more entries than before. Code that asserts a count of formats needs updating.

Same file, different value¶

Four modifications write their Unimod accession in pro_forma (mzLib#1328). In read_records() on a MetaMorpheus .psmtsv or .osmtsv, GG (Ubiquitination Site) is now [UNIMOD:121], both Myristoylation entries [UNIMOD:45] and EQIGG [UNIMOD:846], where they were written by name. Masses are unchanged. A string comparison of pro_forma across the upgrade will see these peptides as different; a pipeline keyed on Unimod will now see ubiquitination sites it silently dropped before.

Same failure, different error¶

A read fault on an existing .mzid (mzLib#1362) is now a BridgeError with error_type MzLibException, naming the file, where it was an IOException. A missing file is still a UsageError. A plain .mzid is now streamed from disk, so reading a large one needs far less memory.
PRIDE errors name the file and the host, never the URL (mzLib#1350), so a reviewer token in a query string cannot reach a log. Every transport failure is still a ServiceUnavailableError. Code that parsed a URL out of an error message will find none.

pyMzLib 0.2.0 (built from mzLib 1.0.592)¶

Files that used to be refused now read¶

mzIdentML (.mzid, and .mzid.gz without unpacking it, mzLib#1313), and MetaMorpheus's AllQuantifiedProteinGroups.tsv and AllQuantifiedPeptides.tsv (mzLib#1347).
Pytheas match output (mzLib#1277). A .txt whose first five lines carry #theoretical_digest reads as Pytheas; any other .txt is still Crux.
Current FlashLFQ and MetaMorpheus QuantifiedPeaks.tsv (mzLib#1345), which have no MBR Score column, read instead of failing with HeaderValidationException.

Same file, different value¶

pro_forma is filled on older MetaMorpheus files (mzLib#1346). MetaMorpheus 1.1.11 and earlier wrote no ProForma column, so read_records() gave None; mzLib now converts full_sequence. Ambiguous (|-joined) or unconvertible rows are still None. Reading is slower: a 295 MB AllPSMs.psmtsv took 21% longer.
mbr_score is None where it was 0 in a FlashLFQ QuantifiedPeaks.tsv (mzLib#1345), for every peak without a score. A 0 used to say "scored, and scored nothing". Seven columns also join the record: organism, peak_fwhm, peak_fwhm_status, pip_q_value, pip_pep, decoy_peptide and random_rt.
MGF one_based_precursor_scan_number is no longer always None (mzLib#1227): mzLib reads it from the PRECURSORSCAN header it writes. It is still None on MGF files mzLib did not write.
FlashLFQ with match-between-runs is reproducible across thread counts (mzLib#1155). Before, more than one thread could give different peptide and protein numbers for identical inputs.

Earlier releases¶

MSFragger retention times are minutes from pyMzLib 0.1.0 (mzLib#1116). Before, an MSFragger psm.tsv passed seconds through, and FlashLFQ quantified it in the wrong window.