We are misusing the string serialization slot

[turbomam]

a string serialization of '{float} {unit}' implies that there are float and unit classes

See also LinkML issue https://github.com/linkml/linkml/issues/674

Switch to LinkML structured patterns

See also

• https://linkml.io/linkml-model/docs/string_serialization/
• https://linkml.io/linkml-model/docs/pattern/ (for native regular expressions)
• structured patterns are most similar to MIxS' Value syntaxes and are used with PatternExpressions: https://linkml.io/linkml-model/docs/PatternExpression/

[turbomam]

Semi-related

Some string serializations are really just lists and could be re-implemented as enumerations

slot	string_serialization
aero_struc	[plane|glider]
built_struc_set	[urban|rural]
ceil_struc	[wood frame|concrete]
contam_screen_input	[reads| contigs]
detec_type	[independent sequence (UViG)|provirus (UpViG)]
fireplace_type	[gas burning|wood burning]
heat_sys_deliv_meth	[conductive|radiant]
host_dependence	[facultative|obligate]
seq_quality_check	[none|manually edited]
shading_device_loc	[exterior|interior]
space_typ_state	[typically occupied|typically unoccupied]
sym_life_cycle_type	[complex life cycle | simple life cycle]
urine_collect_meth	[clean catch|catheter]
wga_amp_appr	[pcr based|mda based]
window_status	[closed|open]

[turbomam]

counts of structured pattern elements in MIxS

string_ser	counts
{text}	239
{float}	90
{unit}	86
{[termID]}	75
{termLabel}	74
{URL}	35
{PMID}	34
{DOI}	34
{integer}	30
{Rn/start_time/end_time/duration}	26
{boolean}	15
{version}	13
{software}	11
{duration}	11
{parameters}	9
{term}	8
{timestamp}	7
{dna}	7
{PMID|DOI|URL}	3
{period}	2
{term label}	2
{NCBI taxid}	2
{rank name}	2
{database}	2
{clustering method}	1
{AF cutoff}	1
{ANI cutoff}	1
{PID}	1
{{text}	1
{day}	1
{term ID}	1
{measurement value}	1
{percentage}	1
{reference}	1
{interval}	1
{has numeric value}	1
{has unit}	1

[turbomam]

Non-alpha characters in the tokens above

Also not including whitespace

count	char	notes
2	_	separates words in a token's name
1	[	literal used with term IDs, like mountain [ENVO:12345678]
1	]	literal used with term IDs, like mountain [ENVO:12345678]
38	{	wraps token. Also, see sieving below
37	}	wraps token
3	/	delimits sub-tokens in {Rn/start_time/end_time/duration}
2	|	delimits alternative tokens for literature references

sieving

{{text}|{float} {unit}};{float} {unit}`

literature references

{PMID|DOI|URL}

[turbomam]

Should clarify the differences between

Canonical

• {[termID]}
• {termLabel}

Variants

• {term ID}, used only by plant_part_maturity
• {term}: geo_loc_name, pos_cont_type, host_of_host_pheno
• {term label}: microb_start, plant_part_maturity

[turbomam]

see also https://github.com/microbiomedata/mixs/pull/37

[ddooley]

Is this all standardized now or is there outstanding work visa vis MIxS or LinkML?

[ramonawalls]

Good quesiton, @ddooley . @turbomam , could you provide an update?