MIxS schema diff: mixs6.0.0 (2022-03-24) to main / 7.0.0-rc1 (2026-07-28)

Source: assets/diff_results/v6.0.0_to_v7.0.0-rc1/schema_comparison_results.yaml (old commit 74744ee, new commit 1591e32).

Four years of change. The raw counts (409 new slots, 553 range changes, 287 class renames) overstate the amount of new content and understate a few things that will actually break consumers. What follows sorts them.

The five changes that explain most of the diff

1. Every class and enum was renamed to CamelCase. 287 of 287 classes and 125 of 125 enums are renamed, not removed. MIGS bacteria -> MigsBa, soil MIMS -> MimsSoil, assembly_qual_enum -> AssemblyQualEnum. Nothing about the content of those classes is in this diff (see Caveats).
2. The quantity value class was deleted and replaced by string plus regex. 173 slots moved from range: quantity value to range: string, and 178 of them gained the same numeric structured pattern (^{scientific_float}( *- *{scientific_float})? *{text}$). 6 more went to float. This, plus similar work elsewhere, accounts for 296 of the 301 pattern changes and all 33 new settings entries (scientific_float, unit, termID, text, and so on) that the structured patterns interpolate.
3. 344 of the 409 "new" slots are generated container slots, not new terms. Everything ending in _data (mims_soil_data, migs_ba_air_data, ...) is a slot on the new MixsCompliantData class, one per non-abstract class. Only 65 genuinely new terms were added.
4. "Package" became "Extension". Old subsets checklist and package are gone, replaced by abstract classes Checklist and Extension; the checklist_package_combination subset became combination_classes; and the 253 combination-class descriptions were rewritten from "Combinatorial checklist X with environmental package Y" to "MIxS Data that comply with the X checklist and the Y Extension".
5. Two new checklists (MISIP variants) and one new extension (Ancient). The 60 new classes are: 4 structural (Checklist, Extension, Ancient, MixsCompliantData), MimarksCMisip plus its 23 extension combinations, MimsMisip plus its 23, and 8 Mims*Ancient combinations. Ancient only combines with MIMS, and only for 8 extensions (host-associated, human-associated, human gut/oral/skin, plant-associated, sediment, soil).

Things that will break or silently change consumers

The term URI base changed. The MIXS prefix went from https://w3id.org/mixs/terms/ to https://w3id.org/mixs/, and the schema id from http://w3id.org/mixs to https://w3id.org/mixs. Every slot_uri in the schema resolves to a different IRI than it did in v6.0.0. This is one line in the diff and the highest-impact item in it.

Two v6.0.0 checklist descriptions were simply wrong and are now fixed. These are class identity corrections, not wording changes:

class	v6.0.0 description	v7 description
MIGS plant -> MigsPl	Minimal Information about a Genome Sequence: plant	...: plasmid
MIGS org -> MigsOrg	...: org	...: organelle
MIGS virus -> MigsVi	...: cultured bacteria/archaea	...: virus
MIMARKS specimen -> MimarksC	Minimal Information about a Marker Specimen	...Marker Sequence
MIMARKS survey -> MimarksS	Minimal Information about a Marker Specimen	...Marker Sequence

Anyone who read "MIGS plant" as the plant-genome checklist (rather than plasmid) was using it incorrectly under v6.0.0.

Two slot descriptions reverse the meaning of the field. Both are size fractionation, both flipped which side of the threshold is excluded:

• size_frac_low: "Materials larger than the size threshold are excluded" -> "Materials smaller than the size threshold are excluded"
• size_frac_up: "Refers to the mesh/pore size used to retain the sample. Materials smaller than the threshold are excluded" -> "Mesh or pore size of the device used to retain the sample. Materials larger than the threshold are excluded"

Data recorded against the v6.0.0 wording of these two slots means the opposite of what v7 says it means.

Added

Classes (60) and subsets (4: environment, investigation, nucleic acid sequence source, sequencing) are described above. The four new subsets replace four of the six removed grouping slots (environment field, investigation field, nucleic acid sequence source field, sequencing field).

Slots: 65 new terms (excluding the 344 *_data containers). They cluster into a small number of themes rather than being scattered:

• Stable isotope probing / isotopolog (14): isotope, isotopolog, isotopolog_approach, isotopolog_atom_frac, isotopolog_atom_pos, isotopolog_dose, isotopolog_incu_time, isotopolog_label, nucleobase_atom_frac, internal_standard, sip_method, plus the density gradient trio gradient_position, gradient_pos_density, gradient_pos_rel_am.
• Ancient DNA / palaeo (14): chrono_age_protocol, chrono_age_remarks, earliest_chrono_age, earliest_chrono_sys, latest_chrono_age, latest_chrono_sys, geological_epoch, cultural_era, stratigraph_context, past_env_broad, past_env_local, palaeopath_status, damage_treatment, host_preserv_state. Matches the new Ancient extension.
• Permits and provenance (11): permit_authority, permit_date, permit_id, permit_scope, biocultural_label, prev_pubs, orig_site_name, orig_site_loc, orig_site_lat, orig_site_lon, context_retrieval_date. This is new territory for MIxS: legal and ethical provenance rather than sample measurement.
• Library prep / capture detail (11): lib_gener_technique, lib_mid_desc, lib_polymerase, lib_strandedness, library_name, capt_pcr_cyc_tot, capt_probe_desc, capt_probe_src_taxid, reamp_pcr_cyc_tot, sop_lib_preparation, sop_experimental.
• Sample handling identifiers (8): batch_ids, samp_alt_lab_ids, samp_category, samp_decont_pretreat, samp_dna_conc, samp_preserv_treatm, nucl_acid_extr_date, nucleic_acid_elution_vol.
• Remainder (7): estimated_genome_size, marker_gene_recov, marker_gene_recov_sw, reads_removed, data_preproc_desc, nose_mouth_teeth_throat_disord, urobiom_sex.

Enums (27 new): AeroStrucEnum, BioCulturalLabelEnum, BuiltStrucSetEnum, CeilStrucEnum, ChronoAgeProtocolEnum, ChronoAgeSysEnum, DamageTreatmentEnum, FireplaceTypeEnum, GeolEpochEnum, HostDependenceEnum, InsdcMissingValueEnum, IsotopeEnum, IsotopologApproachEnum, IsotopologLabelEnum, LibStrandEnum, LibTypeEnum, MoldVisibilityEnum, SampCategoryEnum, SeasonEnum, SeqQualityCheckEnum, ShadingDeviceLocEnum, SpaceTypStateEnum, SymLifeCycleTypeEnum, UrineCollectMethEnum, UrobiomSexEnum, WgaAmpApprEnum, WindowStatusEnum. InsdcMissingValueEnum is worth noting: it brings the INSDC missing-value vocabulary into the schema itself.

Prefixes (7 new): NCIT, SO, chrono, dc, schema, shex, xsd.

Settings (33 new): all of them structured-pattern building blocks (DOI, PMID, URL, scientific_float, unit, termID, termLabel, text, timestamp, duration, software, lat, lon, ...).

Removed

Classes (2): core and quantity value. Both were structural. core was the abstract parent holding the field-grouping slots; quantity value was the value/unit wrapper now replaced by string plus pattern.

Slots (18). Split by why:

• Grouping slots from the deleted core class (6): core field, environment field, investigation field, mixs extension field, nucleic acid sequence source field, sequencing field. Four became subsets; core field and mixs extension field have no successor.
• Slots of the deleted quantity value class (3): has numeric value, has raw value, has unit.
• Genuine term removals (9): estimated_size, salinity_meth, samp_salinity, soil_depth, soil_text_measure, texture_meth, tot_n_meth, tot_phos, url. See "Possible missed renames" below; several of these look like renames rather than deletions.

Enums (27). Only 3 were folded into a surviving enum (door_loc_enum, wall_loc_enum, window_loc_enum -> CompassDirections8Enum, visible only through the slot range changes, not through the rename map). The other 18 dropped enums became unconstrained text, listed under "Cardinality and range changes". The remaining 6 (add_recov_method_enum, assembly_software_enum, compl_score_enum, food_quality_date_enum, pres_animal_insect_enum, samp_purpose_enum) were not referenced as a slot range in v6.0.0 and simply disappear.

Prefixes (2): MIGS, mixs.vocab. Subsets (2): checklist, package (now classes). Imports (25): every per-package module (agriculture, air, ... water). v7 is a single schema file.

Renamed

Classes: all 287, mechanically. Package-prefix word order flipped and CamelCased: soil MIGS bacteria -> MigsBaSoil, hydrocarbon resources-cores MIUVIG -> MiuvigHydrocarbonResourcesCores. Checklist abbreviations regularized: MIMARKS specimen -> MimarksC, MIMARKS survey -> MimarksS, MIGS bacteria -> MigsBa.

Enums: all 125, of which 108 are pure snake_case_enum -> CamelCaseEnum. The other 17 are real consolidations or retitles:

collapsed into	from
DamagedEnum	ceil_cond_enum, floor_cond_enum, int_wall_cond_enum
DamagedRupturedEnum	door_cond_enum, shading_device_cond_enum, window_cond_enum
CompassDirections8Enum	ext_wall_orient_enum, ext_window_orient_enum
GeolAgeEnum	hcr_geol_age_enum, sr_geol_age_enum
SoilHorizonEnum	horizon_enum, soil_horizon_enum
CeilingWallTextureEnum	ceil_texture_enum, wall_texture_enum

plus four retitles: contam_screen_param_enum -> ContamScreenInputEnum, pred_genome_type_enum -> ViralGenomeTypeEnum (and it gained a description, "Types of viral genomes based on Baltimore classification"), single_cell_lysis_appr_enum -> ScLysisApproachEnum, heat_deliv_loc_enum -> HeatSysDelivMethEnum (questionable, see Notes). Net: 125 old enums to 117 new names.

Slots: 26 mappings, but only 10 are renames. The other 16 are merges of near-duplicate slots that both existed in v6.0.0. This is the most useful thing in the slot rename map and it does not read as a merge anywhere else in the diff:

v6.0.0 had both	v7 keeps
sample_name + samp_name	samp_name
samp_collec_method + sample_collec_method + samp_collect_method	samp_collect_method
samp_collec_device + samp_collect_device	samp_collect_device
samp_stor_dur + samp_store_dur	samp_store_dur
samp_stor_loc + samp_store_loc	samp_store_loc
samp_stor_temp + samp_store_temp	samp_store_temp
assembly_quality + assembly_qual	assembly_qual
air particulate matter concentration + air_PM_concen	air_PM_concen
Food_Product_type + food_product_type	food_product_type
Food_source + food_source	food_source
horizon + soil_horizon	soil_horizon
microbial_biomass_meth + micro_biomass_meth	micro_biomass_meth
non_mineral_nutr_regm + non_min_nutr_regm	non_min_nutr_regm
previous_land_use_meth + prev_land_use_meth	prev_land_use_meth
tot_car + tot_carb	tot_carb
host_family_relation + host_family_relationship	host_fam_rel

True renames (10), all abbreviation tightening or prefix fixes: associated resource -> associated_resource, host_infra_specific_name -> host_infra_spec_name, host_infra_specific_rank -> host_infra_spec_rank, single_cell_lysis_appr -> sc_lysis_approach, single_cell_lysis_prot -> sc_lysis_method, texture -> soil_texture, x_16s_recover -> x16s_recover, x_16s_recover_software -> x16s_recover_software.

Subsets (1): checklist_package_combination -> combination_classes.

Possible missed renames

This diff shape has no rename_candidates block, so the following are my own nearest-name matches against the surviving slot set. A maintainer should confirm each and promote the real ones into the tool's rename map, because right now they are reported as removals:

removed	likely successor	confidence
estimated_size	estimated_genome_size (new)	high
texture_meth	soil_texture_meth (existing)	high
soil_text_measure	soil_texture_meth or soil_texture_class (existing)	medium
tot_n_meth	tot_nitro_cont_meth (existing)	medium
tot_phos	tot_phosp or tot_phosphate (both existing)	medium
samp_salinity	salinity (existing)	medium
salinity_meth	no clear successor	low
soil_depth	depth (existing)	low
url	no successor found	low

Of the removed enums, none has a CamelCase counterpart in the new schema, which supports reading them as genuine removals rather than missed renames. The three *_loc_enum cases are the exception: they are real consolidations into CompassDirections8Enum that the rename map does not record.

Cardinality and range changes

Gained required: true (35 slots), where v6.0.0 had none at the slot level: abs_air_humidity, add_recov_method, api, basin, build_occup_type, building_setting, coll_site_geo_feat, collection_date, env_broad_scale, env_local_scale, env_medium, filter_type, geo_loc_name, hc_produced, hcr, heat_cool_type, IFSAC_category, indoor_space, iwf, lat_lon, light_type, occup_density_samp, occup_samp, project_name, rel_air_humidity, samp_collect_point, samp_name, samp_taxon_id, samp_type, seq_meth, space_typ_state, sym_life_cycle_type, typ_occup_density, water_cut. (samp_name is required both under its own name and as the target of the sample_name merge.)

Became multivalued (11): biotic_regm, env_medium, experimental_factor, food_dis_point_city, host_spec_range, solar_irradiance, sop, source_mat_id, ventilation_type, associated_resource (from associated resource), host_fam_rel (from host_family_relationship).

Stopped being multivalued (5): soil_temp, food_clean_proc, farm_water_source, food_product_type, photosynt_activ. These narrow what was previously valid.

Range changes (553 total):

change	count	note
enum renamed	123	naming only; permissible values not compared
quantity value -> string	173	class deleted; 178 slots gained the numeric pattern
quantity value -> float	6	iwf, animal_am_freq, carb_nitro_ratio, season_humidity, surf_humidity, rel_air_humidity
string -> absent	158	no effect; default_range: string is now declared schema-wide
date -> datetime	17	collection_date, cult_isol_date, date_extr_weath, date_last_rain, douche, extreme_event, fertilizer_date, fire, flooding, hrt, iw_bt_date_well, last_clean, menarche, menopause, pregnancy, prod_start_date, cons_purch_date
double -> float	9	all pH slots plus occupancy density
integer -> string	8	number_resident, max_occup, freq_cook, room_occup, exp_pipe, host_occupation, number_plants, number_pets (loosens)
integer/double/quantity value -> absent	5	loosens to string
integer -> float, double -> string	2	occup_samp, timepoint
string -> boolean	6	hysterectomy, twin_sibling, smoker, reassembly_bin, medic_hist_perform, x16s_recover (tightens)
string -> new enum	21	tightens; assembly_qual, contam_screen_input, heat_sys_deliv_meth, season, aero_struc, urine_collect_meth, host_dependence, fireplace_type, built_struc_set, seq_quality_check, shading_device_loc, window_status, ceil_struc, wga_amp_appr, sym_life_cycle_type, space_typ_state, and the 5 *_water_mold slots (ceil, door, shad_dev, wall, window)
string -> float	3	tightens
*_loc_enum -> CompassDirections8Enum	4	door_loc, wall_loc, window_loc, heat_deliv_loc; a consolidation the rename map does not record
enum -> absent (free text)	18	loosens; listed below

The 18 slots that lost their enumeration and are now unconstrained text: cur_land_use, decontam_software, door_type_wood, drug_usage, floor_finish_mat, host_sex, microb_start_count, organism_count, plant_growth_med, room_type, samp_md, source_uvig, special_diet, spikein_count, vis_media, water_source_shared, plus samp_floor and study_complt_stat, which at least gained a regex in exchange. Spot-checking organism_count in the v7 source confirms it now carries only an advisory string_serialization, no range and no pattern. Sixteen previously enumerated fields becoming free text is the largest validation loosening in this release and is not visible in any headline count.

Pattern changes

301 slots changed pattern: 296 added, 4 removed, 1 modified. Note these are LinkML-materialized patterns; in the v7 source most are structured_pattern blocks interpolating the 33 new settings.

Added patterns, by shape:

count	what it constrains
178	numeric value or range, optional units: ^{scientific_float}( *- *{scientific_float})? *{text}$ (the quantity value replacement)
32	PMID / DOI / URL citation form (24 single-valued, 8 a variant)
19	ontology term as label [PREFIX:id]
13	positive integer followed by free text
9	ISO 8601 duration
26	assorted composite forms: term;value unit (7), three semicolon-separated terms (6), term [ID] or integer (5), free text or term [ID] (4), numeric range with unit (3), and singletons including a floor-number pattern for samp_floor

Removed patterns (4), each a loosening: compl_score (^(high|med|low);(0|[0-9]{1,2}|100)%$), contam_screen_param (^(ref db|kmer|coverage|combination);.+), pres_animal_insect (^(cat|dog|rodent|snake|other);\d+$), food_quality_date (^(best by|best if used by|freeze by||use by);YYYY-MM-DD$).

Modified (1): add_recov_method kept its controlled first field and replaced a sprawling ISO 8601 date alternation with a simpler, stricter one.

Title changes

ENA uses MIxS titles as controlled nomenclature, so these are listed individually.

All 287 renamed classes gained a title; none had one in v6.0.0. Titles are generated: MigsBaAgriculture -> "MigsBa combined with Agriculture", Air -> "air", MigsBa -> "MIGS bacteria". Note the combination-class titles are built from the new CamelCase class names, so the human-readable label is now "MigsBa combined with Agriculture" rather than anything resembling the v6.0.0 class name "agriculture MIGS bacteria".

Slot titles changed (12):

slot	v6.0.0 title	v7 title
nose_throat_disord	lung/nose-throat disorder	nose throat disorder
sieving	composite design/sieving	sieving
horizon_meth	soil horizon method	horizon method
samp_taxon_id	Taxonomy ID of DNA sample	taxonomy ID of DNA sample
samp_transport_cont	sample transport  container (double space)	sample transport container
food_trav_vehic	Food shipping transportation vehicle	food shipping transportation vehicle
host_of_host_env_med	host of the symbiotic host environemental medium	host of the symbiotic host environmental medium
air_PM_concen (was air particulate matter concentration)	air_PM_concen	air particulate matter concentration
food_product_type (was Food_Product_type)	Foodon product type	food product type
food_source (was Food_source)	Food source	food source
soil_horizon (was horizon)	horizon	soil horizon
soil_texture (was texture)	texture	soil texture

Four of these (nose_throat_disord, sieving, horizon_meth, food_product_type) change what the title asserts, not just its case. The air_PM_concen row is the interesting one: the merge swapped which of the two duplicate slots supplied the title, so the machine name and the title effectively traded places.

Cosmetic changes (grouped)

• 628 slots dropped a redundant multivalued: false. No semantic effect; false is the LinkML default. This alone is 97% of the 644 multivalued changes.
• 158 slots dropped an explicit range: string. No semantic effect; v7 declares default_range: string schema-wide, which v6.0.0 did not.
• 236 slot descriptions differ only in trailing punctuation or capitalization, almost all of them a stripped trailing period. Example: samp_purpose, "The reason that the sample was collected." -> "The reason that the sample was collected".
• 253 combination-class descriptions were reworded by template from "Combinatorial checklist X with environmental package Y" to "MIxS Data that comply with the X checklist and the Y Extension".
• 22 extension classes replaced a description that just repeated the class name with real prose. Example: air, description "air" -> "A collection of terms appropriate when collecting and sequencing samples obtained from a gaseous environment...".
• 6 slot descriptions are mojibake repairs, non-breaking spaces and degree signs that had been mangled: api, root_med_suppl, root_med_micronutr, root_med_macronutr, root_med_regl, samp_transport_cond.
• 25 slot descriptions are typo or grammar fixes with no change of meaning: animal_am_route (adminstered), diss_oxygen_fluid (oxgen), food_source_age (organim), samp_transport_cont (Conatiner), and 21 others.

That leaves 26 slot descriptions with substantive rewrites, of which the two size-fraction reversals are called out above. The rest are clarifications (alt, depth, pcr_cond, host_body_site, host_body_product, season, store_cond, lat_lon now capping latitude/longitude at 8 decimal places, samp_name rewritten around material-sample identity).

Notes and judgment calls

1. The diff compares only 8 fields: name, description, title, meaning, pattern, range, required, multivalued (src/scripts/diff_two_linkml_mixs_releases.py:171). It does not compare is_a, mixins, class slots lists, slot_usage, enum permissible values, slot_uri, annotations, examples, comments, keywords, or string_serialization. So "125 enums renamed" says nothing about whether their members changed, and the class-level story here is names and descriptions only. A structural comparison is still needed before calling this release reviewed.
2. heat_deliv_loc_enum -> HeatSysDelivMethEnum in the enum rename map looks wrong. The slot heat_deliv_loc actually moved to CompassDirections8Enum, while HeatSysDelivMethEnum is the new range of a different slot, heat_sys_deliv_meth, which was string in v6.0.0. I would treat heat_deliv_loc_enum as consolidated into CompassDirections8Enum alongside door_loc_enum, wall_loc_enum, and window_loc_enum, and drop that map entry.
3. The required: true readings are slot-level only. MIxS expresses most requirement rules per class via slot_usage, which this tool does not compare. The 35 slots above gained a top-level requirement; that is not the same as saying 35 fields became mandatory where they were not before, and it is not the same as saying nothing else changed.
4. The schema itself flags a collision: the new schema's comments say "slot titles that are associated with more than one slot name/SCN: host sex". Worth resolving before release, given the ENA title dependency.
5. main is not a tag. This diff compares mixs6.0.0 to the main branch at commit 1591e32, whose declared version is 7.0.0-rc1. The sibling v6.0.0_to_v7.0.0 folder may point at a different commit; re-run against the release tag before publishing anything from this.
6. I could not classify url (removed slot) or salinity_meth against any surviving slot, and I did not attempt to trace the 6 dropped enums that were never used as a slot range in v6.0.0.