1 GREYLAG TODO LIST -*-outline-*-
4 ==============================================================================
8 1. Replace SEQUEST at SIMR with something at least as good.
9 2. Do better than SEQUEST for things that SIMR cares about.
10 3. Showcase Python w/C-ish inner loop code implementation strategy.
11 4. Try to take the best ideas from other similar programs.
12 5. Greylag as a pedagogical artifact and foundation for further
15 ==============================================================================
20 * Good first impression
22 * Handles at least LCQ input
23 * Nonspecific cleavage only?
24 * Generates SQT output, usable in our pipeline, at least for non-N15 runs
25 * Decent performance/efficiency on our clusters
29 * Basic public website/source release/git archive
30 * Documentation (asciidoc/man pages)
36 * MINI-GOAL: Get something working that can be tested against MM/SEQUEST
37 * MINI-GOAL: basic greylag-process usable on our cluster (no mods)
40 * Need to figure out how to properly merge duplicate peptides within grind
42 There always needs to be a second-best score in order to have a DeltCN!
44 * Implement SQT output program (greylag-sqt)
45 ** R lines to document mass regimes
46 ** A lines for each M line to document regime (index of R) and residue mods
47 ** Try to keep back compatibility by grepping out A/R lines
49 * Compare greylag output to SEQUEST on test-myrimatch example
50 ** also compare against MyriMatch
51 *** do a non-specific run with the existing binary
53 * Try to cut unneeded SWIG types
55 * Compare greylag vs Myrimatch on test-2 case
57 * Design and implement greylag master process (work manifests?)
59 * Evaluate performance differences
61 * Basic test of DTASelect compatibility
63 * Design and implement SQT fileset merger
68 * Look for more dead code to remove
70 * Redo cleavage point code (enzyme and non-specific)
72 = M1 =========================================================================
74 * Implement MM smart +3 model?
77 * Try to generate a valid MyriMatch (bombs on boost random assertion)
79 * Try the MM precursor mass adjustment--much improvement? even a good idea?
81 * Test tolerance monotonicity
85 ** Design similar statistical evaluation
86 ** Look at what we do here (paper)
89 * Add isotope jitter feature, for Orbitrap.
91 xtandem considers one C13 if MH>1000, and one/two C13 if MH>1500. Should we
92 try to predict this based on the peptide sequence? MH probably close
93 enough. What does MyriMatch do?
96 * Implement MyriMatch charge-calling algorithm?
98 * Implement MyriMatch deisotoping?
101 * Figure out how to handle multiple residue mods (delta, isotope, etc)
103 * Clean up [,] (N/C-terminal) mass regime calculations
105 * Pass through the C++ code looking for counts that could conceivably overflow
106 ** Fix or add assertions
108 * Add duplicate peptide masking optimization
109 ** This will obviate the need to detect identical best matches at search time?
110 ** Fix redundant peptide reporting
113 * Time a real mod search vs SEQUEST (and xtandem?), is time reasonable?
114 MAKE SURE PARAMS ARE COMPARABLE! Ballpark correctness?
116 * Create direct DTASelect.txt output?
117 (This seems to be sufficient to support most or all DTASelect output.)
119 * Make a tool to compare greylag vs SEQUEST results by spectrum. Want gross
120 statistics--how many id's are the same, different, missing, etc. For each
121 spectrum, want to see what each program did, and how many times the assigned
122 locus was otherwise id'ed.
124 * Investigate identification differences between greylag and SEQUEST.
126 * Careful timing and correctness check for
127 /n/proteomics/mkc/HsProA-Control_S100_Ti_1_H_2006-03-03_wSHUFFLED-greylag
129 * Design and implement tracing of mass regime/PCA/fixed and non-fixed
130 deltas/etc into output file. Try to stay compatible with xtandem.
132 * PPM error tolerances (MyriMatch doesn't implement this?)
134 * Make --estimate work correctly over cluster. (Currently takes 6 hours to
135 estimate 60--is this worthwhile? Could we simply estimate one bag and
136 multiply by the number of bags??)
138 * Better shuffling than current model.
140 * Useful to scale fragment tolerance by charge, too?
142 * Have --estimate generate a spectrum work count file (*.est?) that can be
143 used by --part-split to generate evenly sized parts. (Check that file is
144 newer than params file and ms2 file arguments, and that all ms2 file
145 arguments were estimated.)
147 * Maybe --part-split should generate a downramp of sizes? It definitely
148 should take into account spectra filtered out (== no work), but this
149 requires reading all spectra before splitting (which takes more time).
151 * Fix "cleavage C-terminal mass change" issue. Should this be interpreted as
152 MONO, ! (first fragment regime), or by regime. Look for similar problems
155 * Make static '[' mod exclude PCA mods.
157 * Code cleanup, especially in new Python code. Maybe put some stuff in
158 classes. Could split into multiple source files.
160 * Mine OMSSA and myrimatch for ideas. Look again at X!Tandem and SEQUEST
163 * Need tool to compare two runs, for regression testing purposes.
165 * Add refinement. (like xtandem?)
167 * Advanced refinement ideas. For example, only search a locus for a hit with N
168 mods if we got a hit for it with 0..N-1 mods (or maybe 0..N-2?). Or, only
169 search a locus non-tryptically (or semi-tryptically) if we got a tryptic hit
172 ** Investigate current SEQUEST search results to see if this looks feasible.
174 * Think about ways to get more id's per hour of processing time.
176 * Try to adapt to instrument accuracy. Maybe start with a narrow parent mass
177 range and adaptively widen it.
179 * Profiling to find slow spots, and for correctness?
181 * Heavy optimization on inner loop. Try running from both ends simultaneously.
183 * Should we try to guarantee that searching is equivalent for all mass regimes,
184 to make comparisons more valid?
186 * Eval speedup: make intensity information integer, or otherwise store it in
187 log form so we can add instead of multiplying?
189 * Try to switch FP code to use integers instead?
191 * Rigorously check all values coming in from Python (at least by assert).
193 * We can now pre-build a peptide index if we want to. The main utility of this
194 is that it would allow us to avoid searching a spectrum against the same
195 peptide multiple times (saving perhaps 30% runtime for one real database).
196 Alternatively, maybe we could just generate a description of peptides to be
199 * Look at moving C++ code to C+ctypes, or maybe pyrex?
201 * Incrementalize the whole program. Want to be able to take an existing run
202 and spend more time on it to get more results, possibly concentrating on a
203 particular kind of modification.
205 * Try to figure out whether SEQUEST is really searching everything, or whether
206 it gives up in certain cases like X!Tandem does.
208 * Isotope S34 and C13 are common (4%, 1%). Is there a good way to look for
209 them? We could look for singleton occurrences pretty cheaply using a delta
212 * Splitting idea: Rather than having all parts be equal, maybe its better for
213 the parts processed first to be bigger, with smallest parts processed last,
214 so that they can fill in the final gaps (leading all processors to finish at
215 about the same time). What should the split curve look like? Linear, but
216 what slope? No split should be smaller than one spectrum (after
219 * Look carefully at the statistics code. Problems?
221 * Implement "cyclic permutation" of xtandem.
223 * Possible generation optimization: Figure out the maximum number of mods,
224 which would be the number that could be added to the smallest, lightest
225 peptide without exceeding the mass of the largest spectrum parent mass.
226 Probably not worth doing? Similarly, if we know all bags of size N are too
227 large, and all deltas are positive, we can skip larger bags.
229 * Can we make the parts restartable? If so, maybe this could be used to load
230 balance, recover from crashes, etc.
232 * Is NOTHROW faster or slower?
234 * If we see a good hit for a spectrum, we could try to see if there's an
235 identifiable tag. If so, could restrict further searching to peptides with
238 * Motif-based differential deltas (like xtandem).
244 * What is this C+57 mod called? Carboxyamidomethyl? +C2H3ON!
246 * Is there anything we can do with neutral losses?
248 * test spectrum synthesis
250 * test semi-tryptic cleavage
252 * double-check handling of FP arithmetic using epsilons (no ==, no strict <)