reads https://tree.opentreeoflife.org/static/statistics/synthesis.json and output of https://gist.github.com/mtholder/7a4161bdfdd82252beb922c77a119d9f

gen_num_studies_data.py

#!/usr/bin/env python
import sys
import json
import datetime
synth_j, unix_ts_and_num_studies = sys.argv[1:]
ts_ns = []
with open(unix_ts_and_num_studies, 'r') as inp:
    for line in inp:
        ls = line.strip().split()
        if ls:
            num_seconds, num_studies = [int(w) for w in ls]
            ts_ns.append([num_seconds, num_studies])
ts_ns.sort()
with open(synth_j, 'r') as inp:
    sd = json.load(inp)

start_time_str = "1970-01-01T00Z"
start_time = datetime.datetime.strptime(start_time_str, "%Y-%m-%dT%HZ")
    
num_synth_trees_by_time = []
for k in sd.keys():
    obj = sd[k]
    num_synth_trees = obj.get('tree_count')
    if num_synth_trees is None:
        num_synth_trees = 484
    x = datetime.datetime.strptime(k, "%Y-%m-%dT%HZ")
    ts = int((x - start_time).total_seconds())
    num_synth_trees_by_time.append((ts, num_synth_trees))
num_synth_trees_by_time.sort()


sys.stdout.write('time\tnum_phyle_studies\tnum_synth_trees\tnum_non_synth_studies\n')
next_synth = num_synth_trees_by_time.pop(0)
next_synth_seconds, next_synth_trees = next_synth
for num_ps_seconds, num_ps_studies in ts_ns:
    if num_ps_seconds > next_synth_seconds:
        try:
            next_synth = num_synth_trees_by_time.pop(0)
            next_synth_seconds, next_synth_trees = next_synth
        except:
            pass
    sys.stdout.write('{}\t{}\t{}\t{}\n'.format(num_ps_seconds, num_ps_studies, next_synth_trees, num_ps_studies - next_synth_trees))